Introduction to Machine Learning
2 min read
Machine Learning is a subset of AI - Artificial Intelligence, that uses statistical algorithms for making accurate or close to accurate predictions over real-time data. There are many other use cases where Machine Learning is used in the real world. We can build machine learning models by training them in either supervised learning or unsupervised learning algorithms and then use these models for predicting and visualizing the statistics for better business development.
Prerequisites for learning machine learning are:
- Python (any high-level programming language to automate the process)
- Sql ( to perform basic operations on datasets)
- Statistics and Linear algebra (basics to understand further topics in ML)
- Machine Learning algorithms 4.1. Linear Regression 4.2. Quadratic Regression 4.3. Logistic Regression 4.4. Semi-Supervised learning and many more. Machine Learning is vast. You can check the complete list of algorithms here.
The basic approach for any problem is to follow the following steps:
- Collect the data (Real-time data is built over years and the ML models accuracy of model predictions is dependent on the amount of data supplied for training the model).
- Cleanse the dataset (Real-time data doesn't have complete information, some information in some rows from a CSV file might be missing. In such cases, to cleanse we could either fill up the missing fields with default values or ignore the entire row, depending upon the requirement)
- Create an ML model using algorithms from libraries like scikit, by supplying a training dataset. (Remember to have some data for testing purposes, as you can check the model accuracy using this testing sub dataset)
- Save the model in lib and check the accuracy. A good ML model is expected to have around 70-90% accuracy. This is not only ideal but realistic.
- If the accuracy isn't that good, try to refine the model. In case of similar results, use other algorithms to find better accuracy in the model.
I have just finished a crash course on youtube from codewithmosh channel. It was great creating a music_recommendation model that predicts genres of music that people liked based on their age and gender. The course used the Linear Regression algorithm from the scikit library for python. It boosted my confidence that I can learn to create machine learning models.
Hoping to learn detailed algorithms and exercise them. I will continue to post my learnings. Hope this was helpful. Thanks for reading. Happy learning!!