MISdata User Manual

MISdata User Manual

MISdata is an R package for retrieving, cleaning, transforming, analyzing, and forecasting stock market data. It provides an end-to-end workflow from index component discovery to ARIMA-based forecasts.

Project status: Experimental, version 0.1.0. Forecasts are statistical estimates and should not be treated as financial advice.

Contents

• Requirements
• Installation
• Quick Start
• Understanding the Data
• 1. Find Stock Symbols
• 2. Download Stock Data
• 3. Clean Missing Values
• 4. Change the Time Period
• 5. Convert the Data Format
• 6. Plot Stock Prices
• 7. Analyze Autocorrelation
• 8. Decompose Seasonality
• 9. Forecast a Stock Series
• Complete Workflow
• Function Reference
• Troubleshooting

Requirements

• R 4.1.0 or later
• An internet connection for downloading index components and stock data
• Access to the package's upstream data sources

The required R package dependencies are installed automatically when MISdata is installed.

Installation

Install the package from GitHub:

install.packages("remotes")
remotes::install_github("MISDataGit/MISdata")

Load it in an R session:

library(MISdata)

To confirm the installed version:

packageVersion("MISdata")

Quick Start

The following example downloads two stocks, cleans the data, plots closing prices, and creates a forecast:

library(MISdata)

prices <- get_stock(
  symbols = c("AAPL", "MSFT"),
  start = "2023-01-01",
  end = "2024-12-31",
  columns = c("Close", "Volume")
)

prices_clean <- clean_stock(prices)

plot_stock(
  prices_clean,
  symbols = c("AAPL", "MSFT"),
  column = "Close",
  format = "long"
)

forecast_result <- forecast_stock(
  prices_clean,
  symbol = "AAPL",
  column = "Close",
  horizon = 30
)

summary(forecast_result$model)
forecast_result$plot

Understanding the Data

get_stock() returns a long-format data.frame. This is the package's main working format.

Column	Meaning	Example
Date	Trading date	2024-01-02
Symbol	Stock ticker	AAPL
Column	Market data field	Close
Value	Numeric observation	185.64

Example:

Date	Symbol	Column	Value
2024-01-02	AAPL	Close	185.64
2024-01-02	AAPL	Volume	82488700
2024-01-02	MSFT	Close	370.87

The available market data fields are:

• Open
• High
• Low
• Close
• Volume
• Adjusted

Run clean_stock() before analysis when the downloaded data contains missing values. The analysis and forecasting functions accept either this long format or the wide xts output created by convert_stock(to = "xts").

1. Find Stock Symbols

Get index components

get_index_components() retrieves the current component list for the Dow Jones Industrial Average or S&P 500 from Wikipedia.

dow_jones <- get_index_components("DJI")
sp500 <- get_index_components("SP500")

head(dow_jones)

The result contains:

• Symbol
• Company
• Industry
• Date_Added

Supported index names are exactly "DJI" and "SP500".

Sample symbols

Use sample_symbols() to select random tickers from an index component table:

symbols <- sample_symbols(
  components = dow_jones,
  n = 3,
  seed = 42
)

symbols

Set seed when the same sample must be reproducible. If n is larger than the number of available components, all available symbols are returned.

You can skip index discovery and provide ticker symbols directly:

symbols <- c("AAPL", "MSFT", "NVDA")

2. Download Stock Data

Use get_stock() to download historical market data:

prices <- get_stock(
  symbols = c("AAPL", "MSFT"),
  start = "2022-01-01",
  end = "2024-12-31",
  columns = c("Open", "High", "Low", "Close", "Volume", "Adjusted"),
  source = "yahoo"
)

Main arguments

Argument	Description	Default
symbols	Character vector of ticker symbols	Required
start	First date in YYYY-MM-DD format	"2020-01-01"
end	Last date in YYYY-MM-DD format	Current date
columns	Market data fields to keep	"OHLCVA"
source	Data source: "yahoo" or "google"	"yahoo"

Yahoo Finance is the recommended default. Data-source availability and ticker coverage are controlled by the upstream service.

If one symbol cannot be downloaded, the function warns and continues with the remaining symbols. It stops if every requested symbol fails.

3. Clean Missing Values

clean_stock() processes each Symbol and Column series independently.

prices_clean <- clean_stock(
  prices,
  na_method = c("trim", "approx")
)

Available methods:

Method	Behavior
"trim"	Removes leading and trailing rows whose values are NA
"approx"	Linearly interpolates missing values inside a series

Both methods are used by default. When both are requested, trimming is applied before interpolation.

Examples:

# Default: trim edges, then interpolate interior gaps
prices_clean <- clean_stock(prices)

# Remove only leading and trailing missing values
prices_trimmed <- clean_stock(prices, na_method = "trim")

# Interpolate only interior gaps
prices_interpolated <- clean_stock(prices, na_method = "approx")

If missing values remain, clean_stock() reports them in a warning. Resolve remaining missing values before calling get_acf(), get_pacf(), plot_seasonal(), or forecast_stock().

4. Change the Time Period

change_period() aggregates a series to monthly, quarterly, or yearly frequency.

monthly <- change_period(prices_clean, period = "monthly")
quarterly <- change_period(prices_clean, period = "quarterly")
yearly <- change_period(prices_clean, period = "yearly")

By default, each OHLCV field is aggregated using financial conventions:

Field	Aggregation
Open	First value in the period
High	Maximum value
Low	Minimum value
Close	Last value in the period
Adjusted	Last value in the period
Volume	Sum

Apply one function to every series by setting aggregate_fn:

monthly_mean <- change_period(
  prices_clean,
  period = "monthly",
  aggregate_fn = mean
)

The output type matches the input type. For example, a long data.frame produces a long data.frame, while an xts input produces an xts result.

5. Convert the Data Format

convert_stock() converts data between common R time-series formats:

prices_xts <- convert_stock(prices_clean, to = "xts")
prices_tsibble <- convert_stock(prices_clean, to = "tsibble")
prices_zoo <- convert_stock(prices_clean, to = "zoo")
prices_ts <- convert_stock(prices_clean, to = "ts")
prices_df <- convert_stock(prices_clean, to = "df")

Supported values for to are:

• "xts"
• "tsibble"
• "zoo"
• "ts"
• "df"

Wide outputs use names such as AAPL_Close and MSFT_Volume.

prices_xts <- convert_stock(
  prices_clean,
  to = "xts",
  format = "wide"
)

A true long converted layout is supported only for to = "tsibble":

prices_long_tsibble <- convert_stock(
  prices_clean,
  to = "tsibble",
  format = "long"
)

For xts, zoo, ts, and df, format = "long" is ignored and a warning is shown. The original output from get_stock() is already a long data.frame.

6. Plot Stock Prices

plot_stock() returns a ggplot object.

One chart with multiple lines

plot_stock(
  prices_clean,
  symbols = c("AAPL", "MSFT"),
  column = "Close",
  format = "wide"
)

One panel per symbol

plot_stock(
  prices_clean,
  symbols = c("AAPL", "MSFT"),
  column = "Close",
  format = "long",
  title = "Closing Prices"
)

Limit the date range

plot_stock(
  prices_clean,
  symbols = "AAPL",
  column = "Close",
  start = "2024-01-01",
  end = "2024-06-30"
)

At most five symbols are plotted. If more are supplied, only the first five are used.

7. Analyze Autocorrelation

get_acf() and get_pacf() calculate simple returns before analyzing autocorrelation:

return = (current value / previous value) - 1

They do not calculate ACF or PACF directly from raw price levels.

ACF

acf_table <- get_acf(
  prices_clean,
  symbols = c("AAPL", "MSFT"),
  column = "Close",
  lag_max = 30
)

head(acf_table)

Request the table and chart together:

acf_result <- get_acf(
  prices_clean,
  symbols = "AAPL",
  plot = TRUE
)

acf_result$table
acf_result$plot

PACF

pacf_result <- get_pacf(
  prices_clean,
  symbols = "AAPL",
  column = "Close",
  lag_max = 30,
  plot = TRUE
)

pacf_result$table
pacf_result$plot

Both functions support at most five symbols and require at least three valid observations per series.

8. Decompose Seasonality

plot_seasonal() uses STL decomposition to display:

• Observed values
• Trend
• Seasonal component
• Remainder

plot_seasonal(
  prices_clean,
  symbol = "AAPL",
  column = "Close"
)

The seasonal period is detected automatically from the average date gap:

Average gap	Period used
Up to 3 days	5
Up to 10 days	52
Up to 45 days	12
Up to 120 days	4
More than 120 days	1

Override the detected period when needed:

plot_seasonal(
  prices_clean,
  symbol = "AAPL",
  column = "Close",
  period = 5,
  start = "2023-01-01",
  end = "2024-12-31"
)

The function requires at least 2 * period observations. Shorter usable windows may produce a warning because decomposition quality can be poor.

9. Forecast a Stock Series

forecast_stock() fits forecast::auto.arima() to one symbol and one field.

result <- forecast_stock(
  prices_clean,
  symbol = "AAPL",
  column = "Close",
  horizon = 30,
  ci_levels = c(80, 95)
)

The returned list contains:

Element	Description
model	Fitted ARIMA model
plot	Historical data, forecast, and confidence bands
forecast	Raw forecast package result

Inspect the result:

summary(result$model)
result$plot
result$forecast$mean
result$forecast$lower
result$forecast$upper

Limit the fitting window:

result <- forecast_stock(
  prices_clean,
  symbol = "AAPL",
  start = "2022-01-01",
  end = "2024-12-31",
  horizon = 60,
  ci_levels = c(50, 80, 95)
)

The function requires at least 10 observations and does not accept missing values. It forecasts one series at a time. Future dates are generated from the observed cadence; daily forecasts skip weekends but do not account for exchange holidays.

Complete Workflow

library(MISdata)

# 1. Get a reproducible sample of Dow Jones symbols
components <- get_index_components("DJI")
symbols <- sample_symbols(components, n = 3, seed = 42)

# 2. Download daily OHLCV data
prices <- get_stock(
  symbols = symbols,
  start = "2022-01-01",
  end = "2024-12-31",
  columns = c("Open", "High", "Low", "Close", "Volume")
)

# 3. Clean each Symbol/Column series
prices_clean <- clean_stock(prices)

# 4. Plot closing prices
price_plot <- plot_stock(
  prices_clean,
  symbols = symbols,
  column = "Close",
  format = "long"
)
price_plot

# 5. Examine return autocorrelation
acf_result <- get_acf(
  prices_clean,
  symbols = symbols,
  column = "Close",
  lag_max = 30,
  plot = TRUE
)
acf_result$plot

# 6. Aggregate to monthly frequency
monthly <- change_period(prices_clean, period = "monthly")

# 7. Plot monthly seasonality for one stock
seasonal_plot <- plot_seasonal(
  monthly,
  symbol = symbols[1],
  column = "Close",
  period = 12
)
seasonal_plot

# 8. Forecast 12 monthly observations
forecast_result <- forecast_stock(
  monthly,
  symbol = symbols[1],
  column = "Close",
  horizon = 12,
  ci_levels = c(80, 95),
  period = 12
)

summary(forecast_result$model)
forecast_result$plot

Function Reference

Function	Purpose
get_index_components()	Retrieve DJI or S&P 500 constituents
sample_symbols()	Randomly select symbols from a component table
get_stock()	Download historical stock data
clean_stock()	Trim and interpolate missing values
change_period()	Aggregate data to monthly, quarterly, or yearly frequency
convert_stock()	Convert between supported time-series formats
plot_stock()	Plot one or more stock series
get_acf()	Calculate and optionally plot return ACF
get_pacf()	Calculate and optionally plot return PACF
plot_seasonal()	Create an STL decomposition plot
forecast_stock()	Fit an ARIMA model and create forecasts

Open the built-in R documentation for full argument details:

?MISdata
?get_stock
?clean_stock
?forecast_stock

Troubleshooting

A symbol cannot be downloaded

Confirm that the ticker is valid for the selected source. Ticker names may differ by exchange, and delisted securities may not be available.

get_stock("AAPL", start = "2024-01-01", source = "yahoo")

All requested symbols fail

Check the internet connection, ticker names, date range, and upstream service availability. get_stock() stops when no requested symbol returns data.

Invalid date error

Use ISO date strings:

start = "2024-01-01"
end = "2024-12-31"

Analysis reports missing values

Clean the long data before analysis:

prices_clean <- clean_stock(prices)
sum(is.na(prices_clean$Value))

No data found for a symbol or column

Inspect the values that are actually present:

unique(prices_clean$Symbol)
unique(prices_clean$Column)

Symbol and column matching is case-sensitive.

STL reports insufficient observations

Use a longer date range or a smaller valid period. STL requires at least 2 * period observations.

Forecasting reports insufficient observations

Use a longer fitting window. forecast_stock() requires at least 10 valid observations after filtering.

More than five symbols are supplied

plot_stock(), get_acf(), and get_pacf() use only the first five symbols. Split larger sets into smaller groups.

Support and License

• Repository: https://github.com/MISDataGit/MISdata
• Issues: https://github.com/MISDataGit/MISdata/issues
• License: GPL (>= 3)