How abhi-105520/sales_prediction works
Main Components
• Data Inputs: Represent the three CSV files as data sources.
• sales.csv – Historical sales data (from 2013 to 2017)
• data_2018.csv & data_2019.csv – New data files for updating predictions
• Data Preprocessing Module: Map the functionality encapsulated in “1_SP_data_preprocessing.ipynb”. This component handles data cleaning, transformation, and extraction of useful information.
• Machine Learning Model Training Module: Represent the “2_SP_model.ipynb” notebook which contains implementations for multiple algorithms (Linear Regression, RandomForest, XGBoost, LSTM Regression).
• Model Evaluation & Comparison Module: Illustrate “3_SP_results.ipynb” where performance of the different models is compared.
• New Data Update Module: Show the “4_SP_newdataupdate.ipynb” notebook which is responsible for uploading and integrating new data (data_2018.csv, data_2019.csv). Stress its irreversible nature as mentioned in the README.
Out Relationships and Data Flow
• From the Data Preprocessing Module, indicate that cleansed and transformed data flows into the Model Training Module.
• Within the Model Training Module, include subcomponents or annotations to denote the four algorithm implementations.
• Direct the results (predictions and model outputs) from the Model Training Module into the Model Evaluation & Comparison Module, indicating a performance analysis workflow.
Architectural Patterns and Design Principles
• The design represents a linear ML pipeline—from raw data processing through model training and evaluation to eventual data update.
• Modularity: each notebook represents a clear component with well-defined responsibilities.
Relevant Technologies and Frameworks
• Since the notebooks are the core of the project, you might note the use of Python and relevant libraries (e.g., scikit-learn, XGBoost, possibly Pandas and NumPy, also libraries for LSTM like TensorFlow or Keras) in your annotations.