Mldlc : Machine Learning Development Life Cycle

hello developers , machine learning is a trending topic we all know but generally we don't know how things work in the industry so for that i will discuss mldc through this article discuss different stages that industries uses while developing a high scale application in machine learning

the stages are which are used in mldc is discuss below

  1. framing the problem
  2. Gathering the data
  3. Data preprocessing
  4. Exploratory data analysis
  5. feature engineering problem
  6. Model training evaluation and selection
  7. model deployment
  8. Testing
  9. Optimization

lets discuss this one by one

framing the problem

framing the problem or understanding the problem is one the most crucial things in whole machine learning we try to understand the problem and try to divide the problem in small segment in this phase we frame the big problem then try to analyze it and divide into more sub problems

Data gathering and data preprocessing

Data is power, data is everything.

Data gathering is also the most important part in mldc cycle because data is like a fuel for machine learning . the more the data the accurate the result .

EXPLORATRY DATA ANALYSIS

exploratry data analysis is the next stage for data analysis in this step we are trying to find hidden insight from the data which is not seen normally in the data . Try to gather and find some hidden pattern in data .

FEATURE ENGINEERING

a worst algorithm with good features column works better than a good algorithm with bad features column

Above is the saying in machine learning industry . feature engineering is an important aspect in machine learning industry because it is step we try to get the best feature for the machine learning model .Feature engineering is an art rather than a skill because every data engineer choose different ways to extract features , every engineer have their signature or unique way to do feature engineering

Model Training Evaluation and algorithm selection

Its is phase where we train model and select algorithm and then evaluate the model for futher processing it is the phase where we train then evaluate and then select the algorithm.

Model deployment

Model deployment is the phase where we deploy the model to the server. there can be two ways to deploy the model to the server it can be offline or batch learning .

OPTIMIZE

Now this is the final phase in the MLDLC here we try to optimize the algorithm for further use case and optimization is also required to save the space and execution in server