Skip to content

Day 3

Machine Learning Development Life Cycle (MLDLC) ⚙️

The MLDLC is a framework that guides the process of building and deploying machine learning models. It is an iterative process, meaning that the steps are not always linear and may need to be repeated as new information is learned.

Steps:

Planning 🎯

  • Define the problem that the machine learning model is intended to solve.
  • Identify the data that will be needed to train the model.
  • Identify the resources that will be available.
  • For example, you are working on house price prediction so you need to know ,at which area you are going to predict, how you getting that data (from kaggle or some APIs etc). How you deal with that problem.

Data Collection 👛

  • After framing / planning the problem, we should start collecting the data from API, kaggle or other methods.

Exploratory Data Analysis 🎫

  • After collecting data, our data is not much cleaner, it is just random facts in a real world scenario.
  • We should most of our time in Exploratory data analysis and data preparation to make machine learning model more accurate.
  • We should analyze each feature and analyze the relationship between every features and to understand on which feature we should work only.
  • Analyze how many missing values are present, are there any outliers etc.

Data preparation 🧹

  • Clean, and prepare the data.
  • Remove outliers.
  • Fill in missing values.
  • Convert the data into a compatible format.

Model engineering (Training & Selection) 🔨

  • Select and train a machine learning algorithm.
  • Compare different algorithms.
  • Choose the one that is best suited for your needs.

Model evaluation 🧪

  • Evaluate the model's performance on a held-out test set using different performance metrics (we'll study in further days).
  • Identify any areas where the model needs improvement.

Model deployment 🚀

  • Integrate the model into an existing software application so that user can interact with our machine learning model.
  • Develop a new application specifically for the model.

Model monitoring 🕵️

  • Monitor the model's performance and store the new data day by day.
  • Make adjustments in model and retrain it as new data appears so it is familiar with new usedcases as well.

Tips

  • Start with a clear understanding of the problem. 💡
  • Use high-quality data. 💎
  • Choose the right algorithm. 🎯
  • Evaluate your model carefully. 🧪
  • Deploy your model in a production environment. 🚀
  • Monitor your model's performance. 🕵️

Conclusion

By following these tips, you can increase your chances of building and deploying successful machine learning models. 💪