LAPTOP PRICE PREDICTOR | USING ML | PTHON | STREAMLIT | HEROKU

 LAPTOP PRICE PREDICTOR USING MACHINE LEARNING  

                                                                        

WELL THE ML MODEL IS MADE & TRAINED ON JUPYTER NOTEBOOK ALSO DID SOME FEATURE ENGINEERING ON THE DATASET, PYCHARM COMMUNITY VERSION(IDE)  WHICH  WAS USED FOR BUILDING THE PYTHON STREAMLET THEN DEPLOYED AT HEROKU SERVER AND THE DATASET ADDED IN THE GITHUB REPOSITORY

    THE MODEL PREDICT THE LAPTOP PRICE ON THE BASIS OF LAPTOP

    • TYPE OF LAPTOP
    • LAPTOP BRANDS
    • SCREENSIZE
    • IPS DISPLAY
    • TOUCHSCREEN OR NOT
    • RAM
    • STORAGE(IN HDD & SDD OR BOTH IS BEING CONSIDERED)
    • GRAPHICS CARD
    • PROCESSOR BRAND
    • OPERATING SYSTEM


     GO THROUGH  AND READ MY "JUPYTER NOTEBOOK" TO GET A OVERALL SCENARIO  OF THE PROJECT

    WHICH ALGORITHMS I TRIED AND WHY?

            I have used THREE algorithms because after analyzing the DATASET I have  found that the Dataset seem Linearly interdependent on other fields so I used the following algorithms

    •  LINEAR REGRESSION
    • GRADIENT BOOSTING
    • RANDOM FOREST REGRESSOR

     1) LINEAR REGRESSION 

    During the Data Analysis, I have found that the price factor linearly depends mostly on specification so I have used Linear Regression Algorithms for building a  MODEL, but this algorithm has not given me an approx result, the R2_score & Mean Absolute Error after training the model was : 
    R2_Score : 0.8073277448418659
    Mean Absolute Error: 0.21017827976428782

     2) GRADIENT BOOSTING  

    After I have tried Linear Regression I thought about Gradient Boosting because this algorithm mathematical R2 is calculated until its matches approx to  the average value of the target variable accepted  likewise Laptop price depending on many factors such as RAM, MEMORY, DISPLAY , this algorithm did it very well the and manage to get 
    R2_Score : 0.885846922907469 
    Mean Absolute Error: 0.15937271818442295

    3) RANDOM FOREST REGRESSOR   

    After using algorithms like Linear Regression & Gradient Boosting I have tried this algorithm because in Random Forest hyperparameters like "max_depth" gives more accurate and approx result for the model, the R2_score & Mean Absolute Error after training the model was : 
    R2_Score : 0.8873402378382488
    Mean Absolute Error: 0.15860130110457718

     DEMO LINK OF THE PROJECT: LAPTOP PRICE PREDICTOR

     GITHUB REPOSITORY: PROJECT LINK  





    Comments

    Post a Comment