When we notice linear relationship pattern in our data we can use Linear model to generate predictions

• We can use this in so many cases np: behavior of the stock market, Sports Performance, business growth
• It can be use to examine the differences between features in data, because some kinds of data fits better than others.
• This type of examination is common in scientific tests e.g: role of statistics in clinical trial in the drug approval

Example 1: Medical charges

I use it on Medical data to predict charges of patients

• I use various graph plots to explore different features in my data and correlation between them
• Then separate smokers and non-smokers in my examinations it is very important to to get the most clear image of the patients
• To boost performance of my model I use PolynomialFeatures to expand predictions possibilities and help my model figure out more patterns in my data
• I make performance measurement of my model using mean squared error and R2 score
• Finally I use hlines plot to see how my Linear Regression model with horizontal line fits in my data

Example 2: Bikes renting

Then I use Linear Regression model to to see which features affect the most on bikes renting frequency

• I plot data distribution to see which feature fits the most to linear pattern
• Then in pure python I create X and Y features of my data and then split them into Train, test and validation sets
• First I train my Linear model on one features which fits the most to the linear pattern
• And get answer that exist some correlation between number of bikes rented and temperature
• Then I use all features and get much higher score, it means that all features merge make more sense for our model predictions
• Next I normalize my data and train Neural Net Regressor model
• And measure performance of both Linear Regression model and Neural Net using MSE
• Final answer for my examination is that in this case Linear Regression Model work better
• I plot both models predictions on Graph

Example 3: Ocean measurements

I use Linear Regression model to examine correlation between temperature and remaining features

• First I figure out size of missing values with help of seaborn and then drop them
• Next I build correlation between all my features and examine particular with help of matplotlib
• Expand data knowledge using new parameter which is density, it help discover correlation between temperature and salinity
• Then I create train and test set using pure python, first for solo salinity and then for all features
• I measure error of my model using Rmse which is MSE squared
• Finally plot my model performance on temperature and salinity data
• And decrease error of my model using all features to train my model Multi Linear Regression

Regression project

In these project i gonna build a model for cancer dataset

First step: Data Preprocessing

I start from data preprocessing, things that will help me are graph plots and functions build in python

• I use: matplotlib and plotly, to build functions i use simply python, sklearn, numpy and math tehniques

Next step: Regression Model