https://github.com/ugurcan222/random-forest-regression-for-time-series-forecasting-with-future-date-predictions

This script performs time series forecasting using Random Forest Regression. It preprocesses data, tunes model parameters with GridSearchCV, and predicts future quantities for specified dates. The results, including actual vs. predicted values and forecasts for future dates, are saved to an Excel file.
https://github.com/ugurcan222/random-forest-regression-for-time-series-forecasting-with-future-date-predictions

Last synced: 4 months ago
JSON representation

This script performs time series forecasting using Random Forest Regression. It preprocesses data, tunes model parameters with GridSearchCV, and predicts future quantities for specified dates. The results, including actual vs. predicted values and forecasts for future dates, are saved to an Excel file.

• Host: GitHub
• URL: https://github.com/ugurcan222/random-forest-regression-for-time-series-forecasting-with-future-date-predictions
• Owner: UgurCan222
• License: mit
• Created: 2024-09-07T13:28:44.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2024-09-07T13:31:56.000Z (9 months ago)
• Last Synced: 2025-01-07T05:43:49.782Z (5 months ago)
• Language: Python
• Size: 209 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Product Quantity Forecasting with Random Forest

This Python script implements a forecasting model to predict product quantities using a Random Forest machine learning model. The entire process involves data preparation, model training, evaluation, and forecasting future quantities. Below is a comprehensive overview of the script's functionality.

## 1. Data Loading and Preparation

- **Load Dataset**: The script begins by loading data from an Excel file located at `C:\Input_Data.xlsx`. This dataset contains historical records of product quantities, dates, locations, and types.

- **Data Conversion**: The 'Date' column is converted from a string format to a datetime format to facilitate time-based analysis. This conversion allows the extraction of additional features.

- **Feature Extraction**: Time-based features are derived from the 'Date' column, including:

  - `DayOfYear`: Day of the year (1-366)

  - `DayOfWeek`: Day of the week (0-6, where Monday=0)

  - `WeekOfYear`: Week number of the year

  - `Month`: Month (1-12)

  - `Year`: Year

**Data Example:**

![Data Example](Data%20Example.png)

## 2. Encoding Categorical Variables

- **Label Encoding**: Categorical variables such as `Location` and `Type` are encoded into numeric values. This encoding is necessary for the machine learning model to process these variables effectively.

- **Mapping Creation**: A dictionary is created to map `ProductCode` to `Type`. This mapping helps in associating each product code with its corresponding type during forecasting.

## 3. Defining Features and Target Variable

- **Feature Selection**: Features (X) used in the model include time-based data and encoded categorical variables such as `DayOfYear`, `DayOfWeek`, `WeekOfYear`, `Month`, `Year`, `Location`, `ProductCode`, and `Type`.

- **Target Variable**: The target variable (y) is the quantity of the product that the model aims to predict.

## 4. Splitting Data for Training and Testing

- **Data Split**: The dataset is divided into training and testing sets using an 80-20 split. This separation allows for model training on one subset and evaluation on another.

## 5. Model Training and Hyperparameter Tuning

- **Model Training**: A Random Forest regression model is trained using the training data.

- **Hyperparameter Tuning**: Grid search is utilized to optimize hyperparameters such as the number of trees (`n_estimators`), maximum depth of each tree (`max_depth`), minimum samples required to split a node (`min_samples_split`), and whether to use bootstrap samples (`bootstrap`).

## 6. Evaluating Model Performance

- **Predictions**: The trained model is used to make predictions on the test set.

- **Performance Metric**: The Root Mean Squared Error (RMSE) is calculated to assess the accuracy of the model's predictions.

**Test Results:**

![Test Results](TestResults(PredictsVsRealDatas)-31July.png)

## 7. Preparing Future Predictions

- **Future Dates**: Future dates are generated to forecast product quantities for the next 7 days.

- **Product Code Distribution**: Historical data is analyzed to understand the distribution of `ProductCode` by day and location. This distribution helps in selecting the most frequent product codes for future forecasting.

## 8. Generating Future Data

- **Future Data Creation**: Data is created for each future date, including product codes and types. The trained model predicts quantities for these future dates.

- **Quantity Prediction**: Predicted quantities are calculated using the trained model, and optional noise is added to simulate variability.

**Predicted Quantities for the Next 7 Days:**

![Predicted Quantities](Predicted7Days.png)

## 9. Finalizing and Saving Results

- **Data Organization**: The results are organized, duplicates are removed.

- **Save to Excel**: The final dataset is sorted by date and location, and then saved to an Excel file at `C:\Forecast.xlsx`.

This script leverages machine learning techniques to forecast future product quantities based on historical data. It enhances planning and decision-making by providing accurate quantity predictions for future dates.

For more details on the implementation, please refer to the `ProjectCode.py` file.