{"id":10478,"date":"2021-06-18T18:22:53","date_gmt":"2021-06-18T12:52:53","guid":{"rendered":"http:\/\/ivyproschool.com\/blog\/?p=10478"},"modified":"2021-06-23T12:06:55","modified_gmt":"2021-06-23T06:36:55","slug":"analyticshala-choose-the-data-analytics-ml-model-in-predictive-analytics-regi-mathew","status":"publish","type":"post","link":"https:\/\/ivyproschool.com\/blog\/analyticshala-choose-the-data-analytics-ml-model-in-predictive-analytics-regi-mathew\/","title":{"rendered":"Analyticshala | Choose the Right Statistical & ML Model in Predictive Analytics | Regi Mathew"},"content":{"rendered":"

Dr. Regi Mathew, a fellowship, PhD holder from IIM Ahmedabad, has extensive experience of 20+ years in the data analytics<\/a> domain. In addition to that, his data science and analytics journey also includes consulting, freelance assignments, and teaching students several topics under the data analytics hood. <\/span>Without any further ado, let\u2019s unearth the important points that he has discussed in the interview, from how he figured out implementing his learning in the fast-changing world and how the analytics boomed.<\/span><\/p>\n

What are the commonly used algorithms? How to choose the best one and when?<\/b><\/h2>\n

Considering the constant changes through which the world is going, Dr Regi suggested that the world is inclined towards Artificial intelligence and Machine learning<\/a>. <\/span>The two most commonly used algorithms are Statistics based algorithms and Machine Learning -based algorithms.\u00a0<\/span>statistics-based is more transparent and gives us the reasoning for explaining the logic behind the end solution. However, ML-based algorithms give predictions that focus on providing much better accuracy, but it does not clearly explain the reason behind the scene. <\/span>To choose the best one according to your situation, you can take your decision well with more pointers to consider through the stats. based approach if you are looking for the reasoning, whereas you can choose ML if you want to focus more on the accuracy score part.\u00a0<\/span><\/p>\n

Which model to choose and how based on the available options?<\/strong><\/h2>\n

Once you have finalized the algorithm, next it comes to which model to choose that will yield the best results. In today’s era, the competition is so tough, and the ML models have become so advanced that the best to choose within the best ML models that you have figured out, for e.g. Random forest, XG boost, etc. and you can go ahead and test them all for selecting the final one based on their accuracy and as per your requirement.<\/span><\/p>\n

What is Bagging and Boosting?<\/strong><\/h2>\n

Bagging stands for bootstrap aggregation, which means we will choose the sample using bootstrap sampling from our data. And, bootstrap sampling means sampling with replacement from the population. In simple words, creating multiple samples from our data and building different models to get their results together to form one algorithm called bagging.<\/span><\/p>\n

Boosting means building a model, and as per the error of the model, we will build another model by changing the weights and bias to minimize the error for better results, and this is a sequential model building process. We tell the algorithm our requirements, and the model building with the best parameters happens in the back end.<\/span><\/p>\n

Which are the best ML algorithms so far?<\/strong><\/h2>\n

The best Machine Learning <\/a>algorithms keep on changing based on the scenario and requirement. However, there are some of the best algorithms until today:<\/span><\/p>\n