100DaysOfMachineLearning

My learning log of these 100 days are here.

You are always welcome to optimize or improve any resource in this repository by following these instructions.

Python Package:

• Numpy: allowing us to work with multidimensional array

• Pandas: to organize data in tabular form and to attach descriptive labels to rows and columns

• Matplotlib: 2D plotting library designed for visualization of numpy computations

• Scipy: tools for mathematics, ML, others

• Seaborn: high-level interface for drawing attractive statistical graphics

• Statsmodels: built on top of numpy and scipy, which integrates with pandas, SM provides good summaries

• Scikit-learn or sklearn: used ML library for below example

How to Save and Load ML Models:

WHAT On various instances, while working on developing a Machine Learning Model, We'll need to save our prediction models to file, and then restore them in order to reuse our previous work to.

WHY We need to save and restore/reload later our ML Model , so as to -

• test our model on/with new data,

• compare multiple models,

• or anything else.

Object serialization: This process / procedure of saving a ML Model is also known as object serialization -

representing an object with a stream of bytes, in order to store it on disk, 
send it over a network or save to a database.

Deserialization: While the restoring/reloading of ML Model procedure is known as deserialization.

Example

from sklearn.externals import joblib
joblib.dump(model, 'filename.pk1')                    #Save in file  
model = joblib.load('filename.pk1')                   #Load from file

Regression

Regression is basically a statistical approach to find the relationship between variables. In machine learning, this is used to predict the outcome of an event based on the relationship between variables obtained from the data-set.

Regression is a task when model attempts to predict continuous values and its evaluation can be done this way.

Linear

1. Simple linear regression

   • How to do simple linear regression
   • Salary Prediction using Simple linear Regression

2. Multiple linear regression

Logistic

* Simple logistic regression
* Multiple logistic regression

Feature Engineering

• Cardinality
• Categorical Encoding
• No Co-linearity
• Normality
• Rare Labels
• One-Hot-Encoding
• Homoscedasticity
• Monotonic

Feature Selection