Kutty Package

Kutty is a Python package developed for simplifying the processes of data cleaning, visualization, and time series analysis. It aims to provide user-friendly interfaces for handling missing values, detecting and treating outliers, creating interactive visualizations, and conducting comprehensive analysis on time series data.

Installation

To install Kutty, ensure you have Python and pip installed on your system. Then, run the following command:

pip install kutty

Features

Kutty is composed of three main modules:

Cleaning Module

• handle_missing_values_advanced: Automatically handles missing values in a DataFrame using various strategies including mean, median, mode, and dropping missing values.
• detect_and_treat_outliers: Detects and treats outliers in a DataFrame using Z-score or IQR methods.

Visualization Module

• create_interactive_scatter_advanced: Generates an interactive scatter plot with extensive customization options.
• create_interactive_time_series: Creates an interactive time series plot, allowing for the highlighting of significant data points.

Time Series Module

• calculate_moving_average: Calculates moving averages, supporting simple, weighted, and exponential types.
• detect_anomalies: Identifies anomalies in time series data using a simple statistical approach.

Usage

Here's how to use the Kutty package in your Python projects:

from kutty.cleaning import handle_missing_values_advanced, detect_and_treat_outliers
from kutty.visualization import create_interactive_scatter_advanced, create_interactive_time_series
from kutty.timeseries import calculate_moving_average, detect_anomalies

import pandas as pd

# Loading your dataset
df = pd.read_csv('your_dataset.csv')

# Cleaning Data
df_cleaned = handle_missing_values_advanced(df, strategy='mean')
df_no_outliers = detect_and_treat_outliers(df_cleaned, method='z_score', threshold=3)

# Visualization
create_interactive_scatter_advanced(df_no_outliers, 'x_column', 'y_column', 'color_column')
create_interactive_time_series(df_no_outliers, 'time_column', 'value_column', title='My Time Series Plot')

# Time Series Analysis
moving_average = calculate_moving_average(df_no_outliers['your_time_series_column'], window=10, type='simple')
anomalies = detect_anomalies(df_no_outliers['your_time_series_column'])

Enhancements and New Features

1. Cleaning Module Enhancements:

   • preprocess_data_for_ml: Function introduced for preprocessing data, including feature scaling and encoding, preparing the dataset for machine learning models.

2. Visualization Module Enhancements:

   • create_multi_variable_plot: Function for creating complex plots that visualize relationships between multiple variables.

3. Time Series Module Enhancements:

   • arima_forecast: A new function is added for time series forecasting using the ARIMA model.
   • prophet_forecast: Function introduced for forecasting using the Prophet model, suitable for datasets with strong seasonal effects.

Usage

Here are brief examples showing how to use the new functionalities:

Preprocess data for machine learning

from kutty.cleaning import preprocess_data_for_ml
df_processed = preprocess_data_for_ml(df)

Create a multi-variable plot

from kutty.visualization import create_multi_variable_plot
create_multi_variable_plot(df, 'category_column', ['feature1', 'feature2', 'feature3'])

Forecasting with ARIMA

from kutty.timeseries import arima_forecast
forecast = arima_forecast(series, order=(1, 1, 1), steps=5)

Forecasting with Prophet

from kutty.timeseries import prophet_forecast
forecast_df = prophet_forecast(df[['ds', 'y']], periods=5, freq='D')

Dependencies

Kutty depends on the following Python libraries: pandas, scipy, numpy, plotly, statsmodels, sklearn, and fbprophet. Ensure these are installed in your environment.

Contributing

Contributions to the Kutty package are welcome! Please feel free to fork the repository, make your changes, and submit a pull request.

License

Kutty is released under the MIT License. See the LICENSE file in the project repository for more details.