Supervised Machine Learning

The entire workflow of any supervised machine learning project can be described in nine steps shown below:

We will divide our every Data science & supervised Machine learning project into these eight simple steps.
You will get the real taste of these steps in future tutorials where we will see each step in practice so don’t worry or stress on them too much for now. Just get an idea of what each step means and in future tutorials i will reference this page alot when we put each and every step in practice.

Define Problem
Before proceeding with any project, you first define the problem you are going to address. Is it a regression problem or a classification one ?

Acquire Data
You acquire your data from different sources. For most of our projects, we will be acquiring our datasets from online sources such as Kaggle

Import Data
We will import our data into python with Pandas

Exploratory Data Analysis
We will explore the data both descriptively and visually here and try to get familar with our data model. Here we will also explore the relations of features with our target.

Data cleaning, Data completing, Feature engineering
Based on the learnings from our ‘Exploratory Data Analysis’ phase, here:

• We will be cleaning our data by dropping the features that are not needed (if there is any)
• We will be completing our data by either filling in the Nan/null values or dropping them altogether
• We may need to create new features based on the existing feature set where required

Get Model
We will get our model and instantiate it

Train or Fit Model
We will train/fit our model with the training dataset

Test Model
We test our model with the test dataset

Model Validation
Since in each project, our target is to predict something. Obviously, in the end, we would like to know, how well did our model do ? In this step, we will validate our model and if the validation results are not satisfactory, we may either retrain our model with different parameters or we will get a different model