direl-ts-tool-kit 0.6.0

A toolbox for time series analysis and visualization.

direl-ts-tool-kit

A Toolbox for Time Series Analysis and Visualization

A lightweight Python library developed to streamline common tasks in time series processing, including data preparation, visualization with a consistent aesthetic style, and handling irregular indices.

Key features and functions

The library provides the following key functionalities, primarily centered around data preparation and plotting.

Data preparation and index management

parse_datetime_index

parse_datetime_index(df_raw, date_column="date", format=None)

Parses a specified column into datetime objects and sets it as the DataFrame index.

This function prepares raw data for time series analysis by ensuring the DataFrame is indexed by the correct datetime type.

generate_dates

generate_dates(df_ts, freq="MS")

Generates a continuous DatetimeIndex covering the time span of the input DataFrame.

The function determines the start and end dates from the existing DataFrame index and creates a new, regular date sequence based on the specified frequency.

reindex_and_aggregate

reindex_and_aggregate(df_ts, column_name, freq="MS")

Re-indexes a time series DataFrame to a regular frequency, aggregates values, and introduces NaN for missing time steps.

This function first identifies the time range from the original (potentially irregular) index, aggregates data if necessary (e.g., if multiple entries exist per time step), and then merges the data onto a complete date range, effectively filling gaps with NaN values.

remove_outliers_by_threshold

remove_outliers_by_threshold(df_ts, column_name, lower_bound, upper_bound)

Replaces values in a specified column with NaN if they fall outside a defined range (outlier removal).

This function identifies data points that are either below the lower bound or above the upper bound and treats them as missing data.

Visualization and styling

plot_time_series

plot_time_series(df_ts, variable, units="", color="BLUE_LINES", time_unit="Year", rot=90, auto_format_label=True)

Plots a time series with custom styling and dual-level grid visibility.

This function automatically sets major and minor time-based locators on the x-axis based on the specified time unit, and formats the y-axis to use scientific notation.

plot_interpolation_analysis

plot_interpolation_analysis(df_original, variable, units="", method="polynomial", order=2, imputation_se=None, time_unit="Year", rot=90)

Performs interpolation on missing data (NaNs) in a specified column and plots the result, highlighting the imputed points with confidence intervals if the Imputation Standard Error (SE) is provided.

save_figure

save_figure(fig, file_name, variable_name="", path="./")

Saves a Matplotlib figure in three common high-quality formats (PNG, PDF, SVG).

The function creates a consistent file name structure: {path}/{file_name}_{variable_name}.{extension}.

heat_map

heat_map(X, y, colors="Blues")

Generates a correlation heatmap plot for a set of features and a target variable.

This function concatenates the feature DataFrame (X) and the target Series (y) to compute and visualize the full pairwise correlation matrix using Seaborn.

pair_plot

pair_plot(X, y)

Generates a cornered pair plot (scatterplot matrix) to visualize relationships between features and the target variable.

The function combines the feature DataFrame (X) and the target Series (y) and uses seaborn.pairplot to create a matrix of scatter plots and histograms. It focuses on the lower triangular part (corner=True) and includes a regression line for trend visualization.

Examples

• Example 1
• Example 2