As discussed in the previous few tutorials, there are various ways in generating the model required for ML.NET.
To recall, there are 3 ways for us to get the model of ML.NET.
- Coding it ourselves
- Using Model Builder Tool
- Using AutoML via CLI to perform model training and picking the best algorithm.
In the previous tutorials, we have gone advance into using AutoML through CLI to generate the model while picking the best algorithm.
It is time to explore into coding them ourselves.
Before we dive into the programming proper, it is good that there are certain steps and procedures to follow.
By the way – one of the key resources which i found helpful is the tutorials on ML.NET docs – https://docs.microsoft.com/en-us/dotnet/machine-learning/tutorials/ Follow step by step through them and you will find various useful materials to leverage on.
Big steps before programming
- Understanding the problem you are solving and choose the tasks involved. ( I will cover this in the next post on how we may want to go about doing this)
- Based on the tasks being chosen, it is then important to prepare the data. Not all data came prepared. You will need to know the features and label of the data as well as what is the input and output class. (I will cover this in the next post on what we need.)
- Based on the prepared data, you should then split the data for testing and training use. This essentially mean to split the data up into testing and training purposes.
- Choose the trainer required in the task. There are various trainers in each tasks, choose the trainer for the task.
- Time to write code