Machine Learning ( Python ) field of Computer Science where we need not explicitly code to build logic and rules. Here the program identifies the pattern from the past available data and predicts. This is a data-driven approach. Actually, we have to collect and build the historical data for a specific task. Using this data we train a model and use this trained model over unseen, real-time, current to future data for prediction. We typically use the machine learning approach if the rule-based approach fails to solve any problem because of the randomness of the outcome. Here we use a probabilistic approach.
Types of Machine Learning :
There are three types of machine learning :
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Here is the full video on Machine Learning and its type.
How to build a Machine Learning Model ( Python ) :
There are several steps that we need to follow to build an AI-ML model. Please go through them stepwise.
-
Data Collection –
If you are building a machine learning model then you need to create a dataset first. To collect data from different data sources you need to retrieve and process data from below data sources.
- Data from Databases ( SQL & NoSQL )
- Data Extraction from PDF, Text File, CSV File.
- Data Extraction from Web
- Data Extraction from Audio File
- Data Extraction from Images and Video.
2. Feature Engineering and Data Preprocessing –
This is a very critical step from an accuracy point of view. In this step, we basically clean, transform, and scale the data. In detail, we perform the below actions.
3. Feature Selection –
If we have multiple features ( variables ) in the dataset and all of them are not significantly impacting the output variable then we should remove them. Multi-feature can also enhance the processing time. To achieve this we perform statistical tests.
4. Model Building –
On the basis of the outcome or task, we choose what algorithm we need to implement. There are multiple tasks we perform usually machine learning.
- 4.1- Classification
- 4.2- Clustering
- 4.3- Regression
- 4.4- Time Series and Forecasting.
After we built up the model we performed two more tasks.
- Model Performance Measurement.
- Hyper Parameter Tuning.
- Model Selection.
5. Serving Model –
This is the final layer to retrieve the prediction from the model. Under this task, we follow the below steps.
5.1 Model Prediction API Development.
5.2 Visualization of the Output.
5.3 Containerization and Model Deployment.
Thanks
Data Science Learner Team
