kamranisg / Digits-Predictor-using-Scikit-Learn

Evaluating 6 ML models to predict digits accurately

Digits-Predictor-using-Scikit-Learn

Description:

We have digits dataset from scikit learn library. Our objective of this project is to compare six supervised machine learning models using the Digits dataset and come out with the best model that accurately predicts the correct digit.

Classifiers:

    1. SUPPORT VECTOR MACHINE
    2. NAIVE BAYES
    3. DECISION TREE
    4. K NEAREST NEIGHBOR
    5. NEURAL NETWORKS
    6. STOCHASTIC GRADIENT DESCCENT

Requirements

Libraries

    1. SCIKIT-LEARN
    2. PANDAS
    3. NUMPY
    4. MATPLOTLIB

System:

    1. PYTHON 3.5+
    2. WINDOWS 10
    3. JUPYTER NOTEBOOK
    4. PIP

Results:

   SUPPORT_VECTOR_MACHINE HAS  8.333333333333332 % ACCURACY

   NAIVE_BAYES HAS  85.27777777777777 % ACCURACY

   DECISION TREE HAS  84.72222222222221 % ACCURACY

   KNEARESTNEIGHBOR HAS  98.61111111111111 % ACCURACY

   NEURAL NETWORKS HAS  42.77777777777778 % ACCURACY

   STOCHASTIC GRADIENT DESCENT HAS  94.72222222222221 % ACCURACY

Our best model is K Nearest Neighbors Model with 98.6% accuray.