Hello there👋, you are probably curious about model building with machine learning. Well, this blog will satisfy your curiosity.

First, it will be good to understand that the whole concept of machine learning comes down to training and testing.

What is predictive modeling?

• Predictive modeling is an aspect of machine learning where models are built to make predictions based on some defined features.

How are these models built?

• Machine learning models are created using some built-in python libraries.

• These data are passed to the model in form of datasets.

• The formats of these datasets are mostly in excel(.xlsx), comma-separated values(.csv), plain text(.txt) etc.

• Every data consists of two parts; features and outcome.

• ML models are built to understand the features of a dataset and their relationship with the outcome.

• Understanding this relationship is called model training

• After training, the model is made to understand real-world features and predict a possible outcome.

• This model can make these predictions, based on what it has learned earlier.

CASE SCENARIO FOR BETTER UNDERSTANDING

With our understanding of the above concept, let's create a case scenario;

Let's create a model that predicts if a boy will eat lunch or not on Saturday;

Let's create some conceptual variables;

m - model to make the prediction

harry - boy which prediction is made based on

l - light food i.e (not an actual meal)

h - heavy food i.e(an actual mean)

n - didn't take anything

On Monday, harry ate h for breakfast, and later that day he took n for lunch,

On Tuesday, harry ate l for breakfast, and later that day he took h for lunch,

On Wednesday, harry ate h for breakfast, and later that day he took l for lunch,

On Thursday, harry ate n for breakfast, and later that day he took h for lunch,

Our model has been given the job to predict what harry will eat for lunch on Friday if he took l for breakfast.

SOLUTION

Using our above concepts that state that every data consists of features and outcomes, in this scenario;

features - (h for breakfast, l for breakfast and n for breakfast)

outcome - lunch

Based on our above concept, our model understands the patterns as follows;

Eating l or n for breakfast, always made him hungry in the afternoon, thereby making him h for lunch.

Eating h for breakfast, never really made him hungry in the afternoon, therefor he always takes either l or n for lunch.

PREDICTION

Based on the patterns, our model will predict that harry will eat h for lunch on Friday.

The above scenario is therefore a simple demonstration of how predictive modeling works.

It would be fair to note that in real-world model building, other features will influence harry's decision e.g fasting, distaste for lunch, food condition etc. Therefore we employ different metrics to guide our model to make predictions.

That will be all for now, hope you got value.

Ensure to like and follow for more exciting content.

Made with ❤️ from rivondave