Notes

Why python for ML?
->Easy, Fast enough,gross plateform,ML libraries
input->data  output->rules for proccessing on data
features :- To identify the labels
feature->ML algorithm(processing on data) -> labels
for data - UCI ML repo
In ML input is features and output are labels
Two types of learning
    1. Supervised learning:
        i.e, Linear regression classifier
        1. already tagged data
        2. features and labels are present
        Two categories:
            1. classification(returns already existing labels)
            2. regression(returns value of new label)
    2. Unsupervised learning
        All machine knows is the data in front of it. No features! No labels!
        1.Data not already tagged
        2. features and labels are not present
        3. training not done
        4. clustering and association
        i.e, Hierarchical clustering

Modules: sklearn, numpy, matplotlib,pandas
sklearn: for datasets
numpy: Array faster than list
matplotlib: to plot graph
pandas: dataframe ,brings data to the RAM for faster access

steps in typical ML problem
    1. problem identification
    2. Data collection and validation
    3. Modal building
    4. Feedback
training and test spliting:
    Given data usage = (To prepare/train modal) + (To test the modal)