Principles of programming

Task 1: ANALYSIS and DESIGN

1. Data Cleaning

• Loaded data on Colab and performed analysis to obtain clean data.

2. Data Visualizations

• Figure 1: Two-by-two relationship analysis of all variables to identify correlations.
• Figure 2: Heat map for correlation analysis to determine relevant parameters.
• Figure 3: Bivariate distribution between salary_in_usd and experience_level.

3. Prediction Model Selection

• Identified the problem as a regression problem through exploratory data analysis (EDA).
• Chose the XGBoost model due to the medium-sized dataset, large number of features, and requirement for high precision in predictions.

4. UML Diagram

• Used two classes: one for data handling and one for tools.
• Data Processing Class:
  • Constants: df for DataFarm, model for the model, and model_fit for the trained model.
  • Methods: doClean (data cleaning), build_chart (create data graphs), train_data (train the model), and forecasts (prediction).
• Tools Class:
  • One method for converting images to base64.

Task 2: IMPLEMENTATION

1. User-Friendly Interface

• Developed a Flask web project for a user-friendly interface.

2. Training Predictive Model

• Loaded data into the model, performed modeling, and evaluated the model's credibility.

3. Testing and Evaluation

• Combined prediction and model training as depicted in the diagram.

4. System Implementation

• Entered parameters to predict salary values.

5. Improvement Suggestions

• Automate address prediction based on user's IP.
• Increase sample size for improved prediction accuracy.
• Enhance input interface aesthetics.
• Add input data accuracy checks.