Skip to content
Go to file


Failed to load latest commit information.
Latest commit message
Commit time
Sep 9, 2019
May 2, 2020
Sep 9, 2019
Apr 26, 2020

CircleCI PyPI Python Versions PyPI Conda ReadTheDocs codecov Downloads

What is it?

D-Tale is the combination of a Flask back-end and a React front-end to bring you an easy way to view & analyze Pandas data structures. It integrates seamlessly with ipython notebooks & python/ipython terminals. Currently this tool supports such Pandas objects as DataFrame, Series, MultiIndex, DatetimeIndex & RangeIndex.


D-Tale was the product of a SAS to Python conversion. What was originally a perl script wrapper on top of SAS's insight function is now a lightweight web client on top of Pandas data structures.

In The News


Related Resources


Where To get It

The source code is currently hosted on GitHub at:

Binary installers for the latest released version are available at the Python package index and on conda using conda-forge.

# conda
conda install dtale -c conda-forge
# if you want to also use "Export to PNG" for charts
conda install -c plotly python-kaleido
# or PyPI
pip install dtale

Getting Started

PyCharm jupyter

Python Terminal

This comes courtesy of PyCharm Feel free to invoke python or ipython directly and use the commands in the screenshot above and it should work

Issues With Windows Firewall

If you run into issues with viewing D-Tale in your browser on Windows please try making Python public under "Allowed Apps" in your Firewall configuration. Here is a nice article: How to Allow Apps to Communicate Through the Windows Firewall

Additional functions available programmatically

import dtale
import pandas as pd

df = pd.DataFrame([dict(a=1,b=2,c=3)])

# Assigning a reference to a running D-Tale process
d =

# Accessing data associated with D-Tale process
tmp =
tmp['d'] = 4

# Altering data associated with D-Tale process
# FYI: this will clear any front-end settings you have at the time for this process (filter, sorts, formatting) = tmp

# Shutting down D-Tale process

# using Python's `webbrowser` package it will try and open your server's default browser to this process

# There is also some helpful metadata about the process
d._data_id  # the process's data identifier
d._url  # the url to access the process

d2 = dtale.get_instance(d._data_id)  # returns a new reference to the instance running at that data_id

dtale.instances()  # prints a list of all ids & urls of running D-Tale sessions

Duplicate data check

To help guard against users loading the same data to D-Tale multiple times and thus eating up precious memory, we have a loose check for duplicate input data. The check runs the following:

  • Are row & column count the same as a previously loaded piece of data?
  • Are the names and order of columns the same as a previously loaded piece of data?

If both these conditions are true then you will be presented with an error and a link to the previously loaded data. Here is an example of how the interaction looks:

Jupyter Notebook

Within any jupyter (ipython) notebook executing a cell like this will display a small instance of D-Tale in the output cell. Here are some examples: assignment instance

If you are running ipython<=5.0 then you also have the ability to adjust the size of your output cell for the most recent instance displayed:

One thing of note is that a lot of the modal popups you see in the standard browser version will now open separate browser windows for spacial convienence:

Column Menus Correlations Describe Column Analysis Instances

JupyterHub w/ Jupyter Server Proxy

JupyterHub has an extension that allows to proxy port for user, JupyterHub Server Proxy

To me it seems like this extension might be the best solution to getting D-Tale running within kubernetes. Here's how to use it:

import pandas as pd

import dtale
import as dtale_app

dtale_app.JUPYTER_SERVER_PROXY = True[1,2,3]))

Notice the command dtale_app.JUPYTER_SERVER_PROXY = True this will make sure that any D-Tale instance will be served with the jupyter server proxy application root prefix:

/user/{jupyter username}/proxy/{dtale instance port}/

One thing to note is that if you try to look at the _main_url of your D-Tale instance in your notebook it will not include the hostname or port:

import pandas as pd

import dtale
import as dtale_app


d =[1,2,3]))
d._main_url # /user/johndoe/proxy/40000/dtale/main/1

This is because it's very hard to promgramatically figure out the host/port that your notebook is running on. So if you want to look at _main_url please be sure to preface it with:

http[s]://[jupyterhub host]:[jupyterhub port]

If for some reason jupyterhub changes their API so that the application root changes you can also override D-Tale's application root by using the app_root parameter to the show() function:

import pandas as pd

import dtale
import as dtale_app[1,2,3]), app_root='/user/johndoe/proxy/40000/`)

Using this parameter will only apply the application root to that specific instance so you would have to include it on every call to show().

JupyterHub w/ Kubernetes

Please read this post

Docker Container

If you have D-Tale installed within your docker container please add the following parameters to your docker run command.

On a Mac: -h `hostname` -p 40000:40000

  • -h, this will allow the hostname (and not the PID of the docker container) to be available when building D-Tale URLs
  • -p, access to port 40000 which is the default port for running D-Tale

On Windows: -p 40000:40000

Everything Else: -h `hostname` --network host

  • -h, this will allow the hostname (and not the PID of the docker container) to be available when building D-Tale URLs
  • --network host, this will allow access to as many ports as needed for running D-Tale processes

Google Colab

This is a hosted notebook site and thanks to Colab's internal function google.colab.output.eval_js & the JS function google.colab.kernel.proexyPort users can run D-Tale within their notebooks.

DISCLAIMER: It is import that you set USE_COLAB to true when using D-Tale within this service. Here is an example:

import pandas as pd

import dtale
import as dtale_app

dtale_app.USE_COLAB = True[1,2,3]))

If this does not work for you try using USE_NGROK which is described in the next section.


This is yet another hosted notebook site and thanks to the work of flask_ngrok users can run D-Tale within their notebooks.

DISCLAIMER: It is import that you set USE_NGROK to true when using D-Tale within this service. Here is an example:

import pandas as pd

import dtale
import as dtale_app

dtale_app.USE_NGROK = True[1,2,3]))

Here are some video tutorials of each:

Service Tutorial Addtl Notes
Google Colab
Kaggle make sure you switch the "Internet" toggle to "On" under settings of your notebook so you can install the egg from pip

It is important to note that using NGROK will limit you to 20 connections per mintue so if you see this error:

Wait a little while and it should allow you to do work again. I am actively working on finding a more sustainable solution similar to what I did for google colab. 🙏

R with Reticulate

I was able to get D-Tale running in R using reticulate. Here is an example:

dtale <- import('dtale')
df <- read.csv('')
dtale$show(df, subprocess=FALSE, open_browser=TRUE)

Now the problem with doing this is that D-Tale is not running as a subprocess so it will block your R console and you'll lose out the following functions:

  • manipulating the state of your data from your R console
  • adding more data to D-Tale

open_browser=TRUE isn't required and won't work if you don't have a default browser installed on your machine. If you don't use that parameter simply copy & paste the URL that gets printed to your console in the browser of your choice.

I'm going to do some more digging on why R doesn't seem to like using python subprocesses (not sure if it something with how reticulate manages the state of python) and post any findings to this thread.

Here's some helpful links for getting setup:


installing python packages

Startup with No Data

It is now possible to run D-Tale with no data loaded up front. So simply call and this will start the application for you and when you go to view it you will be presented with a screen where you can upload either a CSV or TSV file for data.

Once you've loaded a file it will take you directly to the standard data grid comprised of the data from the file you loaded. This might make it easier to use this as an on demand application within a container management system like kubernetes. You start and stop these on demand and you'll be presented with a new instance to load any CSV or TSV file to!


Base CLI options (run dtale --help to see all options available)

Prop Description
--host the name of the host you would like to use (most likely not needed since socket.gethostname() should figure this out)
--port the port you would like to assign to your D-Tale instance
--name an optional name you can assign to your D-Tale instance (this will be displayed in the <title> & Instances popup)
--debug turn on Flask's "debug" mode for your D-Tale instance
--no-reaper flag to turn off auto-reaping subprocess (kill D-Tale instances after an hour of inactivity), good for long-running displays
--open-browser flag to automatically open up your server's default browser to your D-Tale instance
--force flag to force D-Tale to try an kill any pre-existing process at the port you've specified so it can use it

Loading data from arctic(high performance datastore for pandas dataframes) (this requires either installing arctic or dtale[arctic])

dtale --arctic-host mongodb://localhost:27027 --arctic-library jdoe.my_lib --arctic-node my_node --arctic-start 20130101 --arctic-end 20161231

Loading data from CSV

dtale --csv-path /home/jdoe/my_csv.csv --csv-parse_dates date

Loading data from JSON

dtale --json-path /home/jdoe/my_json.json --json-parse_dates date


dtale --json-path http://json-endpoint --json-parse_dates date

Custom Command-line Loaders

Loading data from a Custom loader

  • Using the DTALE_CLI_LOADERS environment variable, specify a path to a location containing some python modules
  • Any python module containing the global variables LOADER_KEY & LOADER_PROPS will be picked up as a custom loader
    • LOADER_KEY: the key that will be associated with your loader. By default you are given arctic & csv (if you use one of these are your key it will override these)
    • LOADER_PROPS: the individual props available to be specified.
      • For example, with arctic we have host, library, node, start & end.
      • If you leave this property as an empty list your loader will be treated as a flag. For example, instead of using all the arctic properties we would simply specify --arctic (this wouldn't work well in arctic's case since it depends on all those properties)
  • You will also need to specify a function with the following signature def find_loader(kwargs) which returns a function that returns a dataframe or None
  • Here is an example of a custom loader:
from dtale.cli.clickutils import get_loader_options

  IMPORTANT!!! This global variable is required for building any customized CLI loader.
  When find loaders on startup it will search for any modules containing the global variable LOADER_KEY.
LOADER_KEY = 'testdata'
LOADER_PROPS = ['rows', 'columns']

def test_data(rows, columns):
    import pandas as pd
    import numpy as np
    import random
    from past.utils import old_div
    from pandas.tseries.offsets import Day
    from dtale.utils import dict_merge
    import string

    now = pd.Timestamp(pd.Timestamp('now').date())
    dates = pd.date_range(now - Day(364), now)
    num_of_securities = max(old_div(rows, len(dates)), 1)  # always have at least one security
    securities = [
        dict(security_id=100000 + sec_id, int_val=random.randint(1, 100000000000),
             str_val=random.choice(string.ascii_letters) * 5)
        for sec_id in range(num_of_securities)
    data = pd.concat([
        pd.DataFrame([dict_merge(dict(date=date), sd) for sd in securities])
        for date in dates
    ], ignore_index=True)[['date', 'security_id', 'int_val', 'str_val']]

    col_names = ['Col{}'.format(c) for c in range(columns)]
    return pd.concat([data, pd.DataFrame(np.random.randn(len(data), columns), columns=col_names)], axis=1)

# IMPORTANT!!! This function is required for building any customized CLI loader.
def find_loader(kwargs):
    test_data_opts = get_loader_options(LOADER_KEY, kwargs)
    if len([f for f in test_data_opts.values() if f]):
        def _testdata_loader():
            return test_data(int(test_data_opts.get('rows', 1000500)), int(test_data_opts.get('columns', 96)))

        return _testdata_loader
    return None

In this example we simplying building a dataframe with some dummy data based on dimensions specified on the command-line:

  • --testdata-rows
  • --testdata-columns

Here's how you would use this loader:

DTALE_CLI_LOADERS=./path_to_loaders bash -c 'dtale --testdata-rows 10 --testdata-columns 5'

Using Swifter

Swifter is a package which will increase performance on any apply() function on a pandas series or dataframe. If install the package in your virtual environment

pip install swifter
# or
pip install dtale[swifter]

It will be used for the following operations:

  • Standard dataframe formatting in the main grid & chart display
  • Column Builders
    • Type Conversions
      • string hex -> int or float
      • int or float -> hex
      • mixed -> boolean
      • int -> timestamp
      • date -> int
    • Similarity Distance Calculation
  • Handling of empty strings when calculating missing counts
  • Building unique values by data type in "Describe" popup

Accessing CLI Loaders in Notebook or Console

I am pleased to announce that all CLI loaders will be available within notebooks & consoles. Here are some examples (the last working if you've installed dtale[arctic]):

  • dtale.show_csv(path='test.csv', parse_dates=['date'])
  • dtale.show_csv(path='http://csv-endpoint', index_col=0)
  • dtale.show_json(path='http://json-endpoint', parse_dates=['date'])
  • dtale.show_json(path='test.json', parse_dates=['date'])
  • dtale.show_arctic(host='host', library='library', node='node', start_date='20200101', end_date='20200101')


Once you have kicked off your D-Tale session please copy & paste the link on the last line of output in your browser

Dimensions/Main Menu

The information in the upper right-hand corner gives grid dimensions

  • lower-left => row count
  • upper-right => column count
  • clicking the triangle displays the menu of standard functions (click outside menu to close it)


The header gives users an idea of what operations have taken place on your data (sorts, filters, hidden columns). These values will be persisted across broswer instances. So if you perform one of these operations and then send a link to one of your colleagues they will see the same thing :)

Notice the "X" icon on the right of each display. Clicking this will remove those operations.

When performing multiple of the same operation the description will become too large to display so the display will truncate the description and if users click it they will be presented with a tooltip where you can crop individual operations. Here are some examples:

Sorts Filters Hidden Columns

Editing Cells

You may edit any cells in your grid (with the exception of the row indexes or headers, the ladder can be edited using the Rename column menu function).

In order to eddit a cell simply double-click on it. This will convert it into a text-input field and you should see a blinking cursor. It is assumed that the value you type in will match the data type of the column you editing. For example:

  • integers -> should be a valid positive or negative integer
  • float -> should be a valid positive or negative float
  • string -> any valid string will do
  • category -> either a pre-existing category or this will create a new category for (so beware!)
  • date, timestamp, timedelta -> should be valid string versions of each
  • boolean -> any string you input will be converted to lowercase and if it equals "true" then it will make the cell True, otherwise False

Users can make use of two protected values as well:

  • "nan" -> numpy.nan
  • "inf" -> numpy.inf

To save your change simply press "Enter" or to cancel your changes press "Esc".

If there is a conversion issue with the value you have entered it will display a popup with the specific exception in question.

Here's a quick demo:

Copy Cells Into Clipboard

Select Copy Paste

One request that I have heard time and time again while working on D-Tale is "it would be great to be able to copy a range of cells into excel". Well here is how that is accomplished:

  1. Shift + Click on a cell
  2. Shift + Click on another cell (this will trigger a popup)
  3. Choose whether you want to include headers in your copy by clicking the checkbox
  4. Click Yes
  5. Go to your excel workbook and execute Ctrl + V or manually choose "Paste"
  • You can also paste this into a standard text editor and what you're left with is tab-delimited data

Main Menu Functions

XArray Operations

  • Convert To XArray: If you have are currently viewing a pandas dataframe in D-Tale you will be given this option to convert your data to an xarray.Dataset. It is as simple as selecting one or many columns as an index and then your dataframe will be converted to a dataset (df.set_index([...]).to_xarray()) which makes toggling between indexes slices much easier.

  • XArray Dimensions: If you are currently viewing data associated with an xarray.Dataset you will be given the ability to toggle which dimension coordinates you're viewing by clicking this button. You can select values for all, some or none (all data - no filter) of your coordinates and the data displayed in your grid will match your selections. Under the hood the code being executed is as follows: ds.sel(dim1=coord1,...).to_dataframe()


View all the columns & their data types as well as individual details of each column

Data Type Display Notes
string If you have less than or equal to 100 unique values they will be displayed at the bottom of your popup
int Anything with standard numeric classifications (min, max, 25%, 50%, 75%) will have a nice boxplot with the mean (if it exists) displayed as an outlier if you look closely.

Outlier Detection

When viewing integer & float columns in the "Describe" popup you will see in the lower right-hand corner a toggle for Uniques & Outliers.

Outliers Filter

If you click the "Outliers" toggle this will load the top 100 outliers in your column based on the following code snippet:

s = df[column]
q1 = s.quantile(0.25)
q3 = s.quantile(0.75)
iqr = q3 - q1
iqr_lower = q1 - 1.5 * iqr
iqr_upper = q3 + 1.5 * iqr
outliers = s[(s < iqr_lower) | (s > iqr_upper)]

If you click on the "Apply outlier filter" link this will add an addtional "outlier" filter for this column which can be removed from the header or the custom filter shown in picture above to the right.

Custom Filter

Apply a custom pandas query to your data (link to pandas documentation included in popup)

Editing Result

You can also see any outlier or column filters you've applied (which will be included in addition to your custom query) and remove them if you'd like.

Context Variables are user-defined values passed in via the context_variables argument to; they can be referenced in filters by prefixing the variable name with '@'.

For example, here is how you can use context variables in a pandas query:

import pandas as pd

df = pd.DataFrame([
  dict(name='Joe', age=7),
  dict(name='Bob', age=23),
  dict(name='Ann', age=45),
  dict(name='Cat', age=88),
two_oldest_ages = df['age'].nlargest(2)
df.query('age in @two_oldest_ages')

And here is how you would pass that context variable to D-Tale:, context_variables=dict(two_oldest_ages=two_oldest_ages))

Here's some nice documentation on the performance of pandas queries

Building Columns

This video shows you how to build the following:

  • Numeric: adding/subtracting two columns or columns with static values
  • Bins: bucketing values using pandas cut & qcut as well as assigning custom labels
  • Dates: retrieving date properties (hour, weekday, month...) as well as conversions (month end)
  • Random: columns of data type (int, float, string & date) populated with random uniformly distributed values.
  • Type Conversion: switch columns from one data type to another, fun. 😄

Summarize Data

This is very powerful functionality which allows users to create a new data from currently loaded data. The operations currently available are:

  • Aggregation: consolidate data by running different aggregations on columns by a specific index
  • Pivot: this is simple wrapper around pandas.Dataframe.pivot and pandas.pivot_table
  • Transpose: transpose your data on a index (be careful dataframes can get very wide if your index has many unique values)
Function Data
No Reshaping


Remove duplicate columns/values from your data as well as extract duplicates out into separate instances.

The folowing screen shots are for a dataframe with the following data:

Function Description Preview
Remove Duplicate Columns Remove any columns that contain the same data as another and you can either keep the first, last or none of these columns that match this criteria. You can test which columns will be removed by clicking the "View Duplicates" button.
Remove Duplicate Column Names Remove any columns with the same name (name comparison is case-insensitive) and you can either keep the first, last or none of these columns that match this criteria. You can test which columns will be removed by clicking the "View Duplicates" button.
Remove Duplicate Rows Remove any rows from your dataframe where the values of a subset of columns are considered duplicates. You can choose to keep the first, last or none of the rows considered duplicated.
Show Duplicates Break any duplicate rows (based on a subset of columns) out into another dataframe viewable in your D-Tale session. You can choose to view all duplicates or select specific groups based on the duplicated value.


Build custom charts based off your data(powered by plotly/dash).

  • The Charts will open in a tab because of the fact there is so much functionality offered there you'll probably want to be able to reference the main grid data in the original tab
  • To build a chart you must pick a value for X & Y inputs which effectively drive what data is along the X & Y axes
    • If you are working with a 3-Dimensional chart (heatmap, 3D Scatter, Surface) you'll need to enter a value for the Z axis as well
  • Once you have entered all the required axes a chart will be built
  • If your data along the x-axis (or combination of x & y in the case of 3D charts) has duplicates you have three options:
    • Specify a group, which will create series for each group
    • Specify an aggregation, you can choose from one of the following: Count, First, Last, Mean, Median, Minimum, Maximum, Standard Deviation, Variance, Mean Absolute Deviation, Product of All Items, Sum, Rolling
      • Specifying a "Rolling" aggregation will also require a Window & a Computation (Correlation, Count, Covariance, Kurtosis, Maximum, Mean, Median, Minimum, Skew, Standard Deviation, Sum or Variance)
      • For heatmaps you will also have access to the "Correlation" aggregation since viewing correlation matrices in heatmaps is very useful. This aggregation is not supported elsewhere
    • Specify both a group & an aggregation
  • You now have the ability to toggle between different chart types: line, bar, pie, wordcloud, heatmap, 3D scatter & surface
  • If you have specified a group then you have the ability between showing all series in one chart and breaking each series out into its own chart "Chart per Group"

Here are some examples:

Chart Type Chart Chart per Group
3D scatter
Maps (Scatter GEO)
Maps (Choropleth)

Y-Axis Toggling

Users now have the ability to toggle between 3 different behaviors for their y-axis display:

  • Default: selecting this option will use the default behavior that comes with plotly for your chart's y-axis
  • Single: this allows users to set the range of all series in your chart to be on the same basis as well as making that basis (min/max) editable
  • Multi: this allows users to give each series its own y-axis and making that axis' range editable

Here's a quick tutorial:

And some screenshots:

Default Single Multi

With a bar chart that only has a single Y-Axis you have the ability to sort the bars based on the values for the Y-Axis

Pre-sort Post-sort

Popup Charts

Viewing multiple charts at once and want to separate one out into its own window or simply move one off to the side so you can work on building another for comparison? Well now you can by clicking the "Popup" button 😄

Copy Link

Want to send what you're looking at to someone else? Simply click the "Copy Link" button and it will save a pre-populated chart URL into your clipboard. As long as your D-Tale process is still running when that link is opened you will see your original chart.

Exporting Charts

You can now export your dash charts (with the exception of Wordclouds) to static HTML files which can be emailed to others or saved down to be viewed at a later time. The best part is that all of the javascript for plotly is embedded in these files so the nice zooming, panning, etc is still available! 💥

Exporting CSV

I've been asked about being able to export the data that is contained within your chart to a CSV for further analysis in tools like Excel. This button makes that possible.


Want to run D-Tale in a jupyter notebook and build a chart that will still be displayed even after your D-Tale process has shutdown? Now you can! Here's an example code snippet show how to use it:

import dtale

def test_data():
    import random
    import pandas as pd
    import numpy as np

    df = pd.DataFrame([
        dict(x=i, y=i % 2)
        for i in range(30)
    rand_data = pd.DataFrame(np.random.randn(len(df), 5), columns=['z{}'.format(j) for j in range(5)])
    return pd.concat([df, rand_data], axis=1)

d =
d.offline_chart(chart_type='bar', x='x', y='z3', agg='sum')

Pro Tip: If generating offline charts in jupyter notebooks and you run out of memory please add the following to your command-line when starting jupyter


Disclaimer: Long Running Chart Requests

If you choose to build a chart that requires a lot of computational resources then it will take some time to run. Based on the way Flask & plotly/dash interact this will block you from performing any other request until it completes. There are two courses of action in this situation:

  1. Restart your jupyter notebook kernel or python console
  2. Open a new D-Tale session on a different port than the current session. You can do that with the following command:, port=[any open port], force=True)

If you miss the legacy (non-plotly/dash) charts, not to worry! They are still available from the link in the upper-right corner, but on for a limited time... Here is the documentation for those: Legacy Charts

Your Feedback is Valuable

This is a very powerful feature with many more features that could be offered (linked subplots, different statistical aggregations, etc...) so please submit issues :)

Network Viewer

This tool gives users the ability to visualize directed graphs. For the screenshots I'll beshowing for this functionality we'll be working off a dataframe with the following data:

Start by selecting columns containing the "To" and "From" values for the nodes in you network and then click "Load":

You can also see instructions on to interact with the network by expanding the directions section by clicking on the header "Network Viewer" at the top. You can also view details about the network provided by the package networkx by clicking the header "Network Analysis".

Select a column containing weighting for the edges of the nodes in the "Weight" column and click "Load":

Select a column containing group information for each node in the "From" column by populating "Group" and then clicking "Load":

Perform shortest path analysis by doing a Shift+Click on two nodes:

View direct descendants of each node by clicking on it:

You can zoom in on nodes by double-clicking and zoom back out by pressing "Esc".


Shows a pearson correlation matrix of all numeric columns against all other numeric columns

  • By default, it will show a grid of pearson correlations (filtering available by using drop-down see 2nd table of screenshots)
  • If you have a date-type column, you can click an individual cell and see a timeseries of pearson correlations for that column combination
    • Currently if you have multiple date-type columns you will have the ability to toggle between them by way of a drop-down
  • Furthermore, you can click on individual points in the timeseries to view the scatter plot of the points going into that correlation
    • Within the scatter plot section you can also view the details of the PPS for those data points in the chart by hovering over the number next to "PPS"
Matrix PPS Timeseries Scatter
Col1 Filtered Col2 Filtered Col1 & Col2 Filtered

When the data being viewed in D-Tale has date or timestamp columns but for each date/timestamp vlaue there is only one row of data the behavior of the Correlations popup is a little different

  • Instead of a timeseries correlation chart the user is given a rolling correlation chart which can have the window (default: 10) altered
  • The scatter chart will be created when a user clicks on a point in the rollign correlation chart. The data displayed in the scatter will be for the ranges of dates involved in the rolling correlation for that date.
Data Correlations

Predictive Power Score

Predictive Power Score (using the package ppscore) is an asymmetric, data-type-agnostic score that can detect linear or non-linear relationships between two columns. The score ranges from 0 (no predictive power) to 1 (perfect predictive power). It can be used as an alternative to the correlation (matrix). WARNING: This could take a while to load.

This page works similar to the Correlations page but uses the PPS calcuation to populate the grid and by clicking on cells you can view the details of the PPS for those two columns in question.


Heat Map

This will hide any non-float or non-int columns (with the exception of the index on the right) and apply a color to the background of each cell.

  • Each float is renormalized to be a value between 0 and 1.0
  • You have two options for the renormalization
    • By Col: each value is calculated based on the min/max of its column
    • Overall: each value is caluclated by the overall min/max of all the non-hidden float/int columns in the dataset
  • Each renormalized value is passed to a color scale of red(0) - yellow(0.5) - green(1.0)

Turn off Heat Map by clicking menu option you previously selected one more time

Highlight Dtypes

This is a quick way to check and see if your data has been categorized correctly. By clicking this menu option it will assign a specific background color to each column of a specific data type.

category timedelta float int date string bool
purple orange green light blue pink white yellow

Highlight Missing

  • Any cells which contain nan values will be highlighted in yellow.
  • Any string column cells which are empty strings or strings consisting only of spaces will be highlighted in orange.
  • will be prepended to any column header which contains missing values.

Highlight Outliers

Highlight any cells for numeric columns which surpass the upper or lower bounds of a custom outlier computation.

  • Lower bounds outliers will be on a red scale, where the darker reds will be near the maximum value for the column.
  • Upper bounds outliers will be on a blue scale, where the darker blues will be closer to the minimum value for the column.
  • will be prepended to any column header which contains outliers.

Highlight Range

Highlight any range of numeric cells based on three different criteria:

  • equals
  • greater than
  • less than

You can activate as many of these criteria as you'd like nad they will be treated as an "or" expression. For example, (x == 0) or (x < -1) or (x > 1)

Selections Output

Low Variance Flag

Show flags on column headers where both these conditions are true:

  • Count of unique values / column size < 10%
  • Count of most common value / Count of second most common value > 20

Here's an example of what this will look like when you apply it: |

Code Exports

Code Exports are small snippets of code representing the current state of the grid you're viewing including things like:

  • columns built
  • filtering
  • sorting

Other code exports available are:

  • Column Analysis
  • Correlations (grid, timeseries chart & scatter chart)
  • Describe
  • Charts built using the Chart Builder

Type Code Export
Main Grid
Correlation Grid
Correlation Timeseries
Correlation Scatter

Export CSV

Export your current data to either a CSV or TSV file:

Load Data & Sample Datasets

So either when starting D-Tale with no pre-loaded data or after you've already loaded some data you now have the ability to load data or choose from some sample datasets directly from the GUI:

Here's the options at you disposal:

  • Load a CSV/TSV file by dragging a file to the dropzone in the top or select a file by clicking the dropzone
  • Load a CSV/TSV or JSON directly from the web by entering a URL (also throw in a proxy if you are using one)
  • Choose from one of our sample datasets:
    • US COVID-19 data from NY Times (updated daily)
    • Script breakdowns of popular shows Seinfeld & The Simpsons
    • Movie dataset containing release date, director, actors, box office, reviews...
    • Video games and their sales
    • pandas.util.testing.makeTimeDataFrame


This will give you information about other D-Tale instances are running under your current Python process.

For example, if you ran the following script:

import pandas as pd
import dtale[dict(foo=1, bar=2, biz=3, baz=4, snoopy_D_O_double_gizzle=5)]))[
    dict(a=1, b=2, c=3, d=4),
    dict(a=2, b=3, c=4, d=5),
    dict(a=3, b=4, c=5, d=6),
    dict(a=4, b=5, c=6, d=7)
]))[range(6), range(6), range(6), range(6), range(6), range(6)]), name="foo")

This will make the Instances button available in all 3 of these D-Tale instances. Clicking that button while in the first instance invoked above will give you this popup:

The grid above contains the following information:

  • Process: timestamp when the process was started along with the name (if specified in
  • Rows: number of rows
  • Columns: number of columns
  • Column Names: comma-separated string of column names (only first 30 characters, hover for full listing)
  • Preview: this button is available any of the non-current instances. Clicking this will bring up left-most 5X5 grid information for that instance
  • The row highlighted in green signifys the current D-Tale instance
  • Any other row can be clicked to switch to that D-Tale instance

Here is an example of clicking the "Preview" button:


This will give you information about what version of D-Tale you're running as well as if its out of date to whats on PyPi.

Up To Date Out Of Date

Refresh Widths

Mostly a fail-safe in the event that your columns are no longer lining up. Click this and should fix that


Toggle between light & dark themes for your viewing pleasure (only affects grid, not popups or charts).

Light Dark

Reload Data

Force a reload of the data from the server for the current rows being viewing in the grid by clicking this button. This can be helpful when viewing the grid from within another application like jupyter or nested within another website.


Pretty self-explanatory, kills your D-Tale session (there is also an auto-kill process that will kill your D-Tale after an hour of inactivity)

Column Menu Functions


These interactive filters come in 3 different types: String, Numeric & Date. Note that you will still have the ability to apply custom filters from the "Filter" popup on the main menu, but it will get applied in addition to any column filters.

Type Filter Data Types Features
String strings & booleans The ability to select multiple values based on what exists in the column. Notice the "Show Missing Only" toggle, this will only show up if your column has nan values
Date dates Specify a range of dates to filter on based on start & end inputs
Numeric ints & floats For integers the "=" will be similar to strings where you can select multiple values based on what exists in the column. You also have access to other operands: <,>,<=,>=,() - "Range exclusve", [] - "Range inclusive".

Moving Columns

All column movements are saved on the server so refreshing your browser won't lose them 👌

Hiding Columns

All column movements are saved on the server so refreshing your browser won't lose them 👌


As simple as it sounds, click this button to delete this column from your dataframe.


Update the name of any column in your dataframe to a name that is not currently in use by your dataframe.


This feature allows users to replace content on their column directly or for safer purposes in a brand new column. Here are the options you have:

Type Data Types Description Menu
Value(s) all Replace specific values in a column with raw values, output from another column or an aggregation on your column
Spaces Only strings Replace string values consisting only of spaces with raw values
Contains Char/Substring strings Replace string values containing a specific character or substring
Scikit-Learn Imputer numeric Replace missing values with the output of using different Scikit-Learn imputers like iterative, knn & simple

Here's a quick demo:


Adds your column to "locked" columns

  • "locked" means that if you scroll horizontally these columns will stay pinned to the right-hand side
  • this is handy when you want to keep track of which date or security_id you're looking at
  • by default, any index columns on the data passed to D-Tale will be locked


Removed column from "locked" columns


Applies/removes sorting (Ascending/Descending/Clear) to the column selected

Important: as you add sorts they sort added will be added to the end of the multi-sort. For example:

Action Sort
click "a"
sort asc a (asc)
click "b" a (asc)
sort desc a (asc), b(desc)
click "a" a (asc), b(desc)
sort None b(desc)
sort desc b(desc), a(desc)
click "X" on sort display


Apply simple formats to numeric values in your grid

Type Editing Result

For all data types you have the ability to change what string is ued for display.

For numbers here's a grid of all the formats available with -123456.789 as input:

Format Output
Precision (6) -123456.789000
Thousands Sep -123,456.789
Abbreviate -123k
Exponent -1e+5
BPS -1234567890BPS
Red Negatives -123457

For strings you can apply the follwoing formats:

  • Truncation: truncate long strings to a certain number of characters and replace with an allipses "..." and see the whole value on hover.
  • Hyperlinks: If your column is comprised of URL strings you can make them hyperlinks which will open a new tab

Column Analysis

Based on the data type of a column different charts will be shown.

Chart Data Types Sample
Histogram Float, Int
Value Counts Int, String, Bool, Date, Category
Word Value Counts String
Category Float
Geolocation* Int, Float

Histogram can be displayed in any number of bins (default: 20), simply type a new integer value in the bins input

Value Count by default, show the top 100 values ranked by frequency. If you would like to show the least frequent values simply make your number negative (-10 => 10 least frequent value)

Value Count w/ Ordinal you can also apply an ordinal to your Value Count chart by selecting a column (of type int or float) and applying an aggregation (default: sum) to it (sum, mean, etc...) this column will be grouped by the column you're analyzing and the value produced by the aggregation will be used to sort your bars and also displayed in a line. Here's an example:

Word Value Count you can analyze string data by splitting each record by spaces to see the counts of each word. This chart has all the same functions available as "Value Counts". In addition, you can select multiple "Cleaner" functions to be applied to your column before building word values. These functions will perform operations like removing punctuation, removing numeric character & normalizing accent characters.

Category (Category Breakdown) when viewing float columns you can also see them broken down by a categorical column (string, date, int, etc...). This means that when you select a category column this will then display the frequency of each category in a line as well as bars based on the float column you're analyzing grouped by that category and computed by your aggregation (default: mean).

Geolocation when your data contains latitude & longitude then you can view the coverage in a plotly scattergeo map. In order to have access this chart your dataframe must have at least one of each of these types of columns:

  • "lat" must be contained within the lower-cased version of the column name and values be between -90 & 90
  • "lon" must be contained within the lower-cased version of the column name and values be between -180 & 180


These are key combinations you can use in place of clicking actual buttons to save a little time:

Keymap Action
shift+m Opens main menu*
shift+d Opens "Describe" page*
shift+f Opens "Custom Filter"*
shift+b Opens "Build Column"*
shift+c Opens "Charts" page*
shift+x Opens "Code Export"*
esc Closes any open modal window & exits cell editing

* Does not fire if user is actively editing a cell.

Menu Functions Depending on Browser Dimensions

Depending on the dimensions of your browser window the following buttons will not open modals, but rather separate browser windows: Correlations, Describe & Instances (see images from Jupyter Notebook, also Charts will always open in a separate browser window)

For Developers


Clone the code (git clone ssh://, then start the backend server:

$ git clone ssh://
# install the dependencies
$ python develop
# start the server
$ python dtale --csv-path /home/jdoe/my_csv.csv --csv-parse_dates date

You can also run dtale from PyDev directly.

You will also want to import javascript dependencies and build the source:

$ npm install
# 1) a persistent server that serves the latest JS:
$ npm run watch
# 2) or one-off build:
$ npm run build

Running tests

The usual npm test command works:

$ npm test

You can run individual test files:

$ TEST=static/__tests__/dtale/DataViewer-base-test.jsx npm run test-file


You can lint all the JS and CSS to confirm there's nothing obviously wrong with it:

$ npm run lint -s

You can also lint individual JS files:

$ npm run lint-js-file -s -- static/dtale/DataViewer.jsx

Formatting JS

You can auto-format code as follows:

$ npm run format

Docker Development

You can build python 27-3 & run D-Tale as follows:

$ yarn run build
$ docker-compose build dtale_2_7
$ docker run -it --network host dtale_2_7:latest
$ python
>>> import pandas as pd
>>> df = pd.DataFrame([dict(a=1,b=2,c=3)])
>>> import dtale

Then view your D-Tale instance in your browser using the link that gets printed

You can build python 36-1 & run D-Tale as follows:

$ yarn run build
$ docker-compose build dtale_3_6
$ docker run -it --network host dtale_3_6:latest
$ python
>>> import pandas as pd
>>> df = pd.DataFrame([dict(a=1,b=2,c=3)])
>>> import dtale

Then view your D-Tale instance in your browser using the link that gets printed

Global State/Data Storage

If D-Tale is running in an environment with multiple python processes (ex: on a web server running gunicorn) it will most likely encounter issues with inconsistent state. Developers can fix this by configuring the system D-Tale uses for storing data. Detailed documentation is available here: Data Storage and managing Global State

Startup Behavior

Here's a little background on how the function works:

  • by default it will look for ports between 40000 & 49000, but you can change that range by specifying the environment variables DTALE_MIN_PORT & DTALE_MAX_PORT
  • think of sessions as python consoles or jupyter notebooks
  1. Session 1 executes our state is:
Session Port Active Data IDs URL(s)
1 40000 1 http://localhost:40000/dtale/main/1
  1. Session 1 executes our state is:
Session Port Active Data IDs URL(s)
1 40000 1,2 http://localhost:40000/dtale/main/[1,2]
  1. Session 2 executes our state is:
Session Port Active Data IDs URL(s)
1 40000 1,2 http://localhost:40000/dtale/main/[1,2]
2 40001 1 http://localhost:40001/dtale/main/1
  1. Session 1 executes, port=40001, force=True) our state is:
Session Port Active Data IDs URL(s)
1 40001 1,2,3 http://localhost:40001/dtale/main/[1,2,3]
  1. Session 3 executes our state is:
Session Port Active Data IDs URL(s)
1 40001 1,2,3 http://localhost:40001/dtale/main/[1,2,3]
3 40000 1 http://localhost:40000/dtale/main/1
  1. Session 2 executes our state is:
Session Port Active Data IDs URL(s)
1 40001 1,2,3 http://localhost:40001/dtale/main/[1,2,3]
3 40000 1 http://localhost:40000/dtale/main/1
2 40002 1 http://localhost:40002/dtale/main/1
  1. Session 4 executes, port=8080) our state is:
Session Port Active Data IDs URL(s)
1 40001 1,2,3 http://localhost:40001/dtale/main/[1,2,3]
3 40000 1 http://localhost:40000/dtale/main/1
2 40002 1 http://localhost:40002/dtale/main/1
4 8080 1 http://localhost:8080/dtale/main/1
  1. Session 1 executes dtale.get_instance(1).kill() our state is:
Session Port Active Data IDs URL(s)
1 40001 2,3 http://localhost:40001/dtale/main/[2,3]
3 40000 1 http://localhost:40000/dtale/main/1
2 40002 1 http://localhost:40002/dtale/main/1
4 8080 1 http://localhost:8080/dtale/main/1
  1. Session 5 sets DTALE_MIN_RANGE to 30000 and DTALE_MAX_RANGE 39000 and executes our state is:
Session Port Active Data ID(s) URL(s)
1 40001 2,3 http://localhost:40001/dtale/main/[2,3]
3 40000 1 http://localhost:40000/dtale/main/1
2 40002 1 http://localhost:40002/dtale/main/1
4 8080 1 http://localhost:8080/dtale/main/1
5 30000 1 http://localhost:30000/dtale/main/1


Have a look at the detailed documentation.


  • Back-end
    • dash
    • dash_daq
    • Flask
    • Flask-Compress
    • flask-ngrok
    • Pandas
    • plotly
    • scikit-learn
    • scipy
    • xarray
    • arctic [extra]
    • redis [extra]
    • rpy2 [extra]
  • Front-end
    • react-virtualized
    • chart.js


D-Tale has been under active development at Man Numeric since 2019.

Original concept and implementation: Andrew Schonfeld


Contributions welcome!


D-Tale is licensed under the GNU LGPL v2.1. A copy of which is included in LICENSE