Iris Data Analysis

Chin Hock Yang

About Project

This project aims to perform data analysis on Iris flower classes in a garden and to create a program that is capable of making predictions of a new Iris flower. The project repository contains of 2 main files, analysis.ipynb and app.py.

analysis.ipynb

This Jupyter Notebook assesses the quality of the Iris data, and constructs Supervised Machine Learning models such as Decision Tree, Logistic Regression and K-Nearest Neighbours models that are capable of making predictions on a new Iris flower data. It also contains codes that perform Unsupervised Learning such as identifying data clusters using K-Means Clustering and using Cosine-Similarity to measure the "similarity" between data points. A Logistic Regression model is generated and stored in the model folder within this notebook, which will be used by the programme created in app.py.

app.py

This Python script contains a setup of a Plotly Dash dashboard application. Presently the application is best-fitted for usage on Google Chrome browser on a laptop or computer monitor display. There are 3 main sections in this application.

Section	Description
Data Exploration	View the proportion of the Iris classes, and inspect for missing data in the dataset
Comparison across Iris Types	Visualise the distribution and differences of the various attributes and characteristics of the Iris types
New Data Prediction	Provide value of the attributes of the new Iris data, and view the predicted Iris type and 10 most similar existing Iris data to the input Iris data

Data Exploration

Comparison Across Iris Types

New Data Species Prediction

New Data Species Prediction

New Data Comparison to Existing Data

Environment Setup

This project uses local Virtual Environment to manage project dependencies. (environment creation and activation guide). To setup the repository, Python (version 3.8 and above) and Pip (package installer) will need to be first installed on the local machine.

Steps to setup Virtual Environment

1. In the project root directory, run python -m venv venv to create a virtual environment folder venv at the root level of the repository
2. Activate the environment by running source venv/bin/activate (for Mac Users) or venv\Scripts\activate (for Windows Users)
3. Run the command pip install -r requirements.txt after activating virtual environment to install all necessary dependencies.

Getting Started

If no data are present in the data folder

1. Download the data iris.data from the data source and rename the file as iris.csv. This file should contain 5 columns of data and should not contain any header rows.
2. Store the iris.csv in the data folder.
3. If the name of the main dataset is changed, the filename within pd.read_csv function in Section 1.2 of analysis.ipynb and the IRIS_CSV_DATA_FILENAME variable in app.py will need to be updated accordingly.

If no files are present in the model folder

1. Run all the cells in the analysis.ipynb Jupyter Notebook to generate a best_model.pkl file (containing a fitted Logistic Regression model)
2. Within the notebook, changes and improvements can be made to the model and the model can be saved into the model folder using the save_file function in the notebook.
3. If the name of the main predictive model is changed, the filename within load_file function in Section 6.3 of analysis.ipynb and the MODEL_PKL_FILENAME variable in app.py will need to be updated accordingly.

If the data and model files are already present in their respective folders folder

1. Activate the virtual environment in the terminal or command prompt
2. Run python app.py to start the Dashboard
3. The app should be running on http://localhost:8050/