[FEAT]: Creation of ML model regarding win prediction of our GC cricket tournament.

[geekofycoder]

So the idea is that we have seen the win probability of google during the IPL and international matches. How about we create a same system but its for our own college and GC!

[zakhaev26]

Nice idea.Have you thought anything about the implementation?
Or we can roll about it here
CC: @majorbruteforce @punitkr03

[geekofycoder]

Yes absolutely.The implementation would be using the ML algos but I was asking GPT and it suggested Kafka for streaming services.Can you brief or explain how Kafka helps in real time analysis

[geekofycoder]

For ML algo Random Forest is something that is coming to my mind and also suggests by GPT and some videos. Suggestions for optimisation?

[zakhaev26]

Suggestions for optimisation?

Currently the ones who are working on this project ain't having any idea on ML :p

I would recommend to bring Anshuman (Sophomore,IT) to include in this thread/anyone from PSoc/ML to discuss on this as they have greater knowledge on ML stuff.

[punitkr03]

Look for a way to implement random forest ML algorithm. It fiits our use-case well. Also it is less prone to overfitting which will be more accurate in less amount of training data. @geekofycoder

[zakhaev26]

Yes absolutely.The implementation would be using the ML algos but I was asking GPT and it suggested Kafka for streaming services.Can you brief or explain how Kafka helps in real time analysis

Kafka is a very complex thing at large scale,for eg if you have a huge amount of data to be processed by your backend servers/ML models/any resource intensive worker,Handling data at scales of Millions/sec would be hard,for eg,your db would be massacred due to insertion of so much throughputs of data in such a short time.
Kafka helps by managing this through using a Queue / DLQ based Publication-Subscription model that helps in distributing the workload and allowing things to work without going down as a response to so much data ingestion at once.

This is a very small idea on Kafka,but I would suggest to watch this video that have a real world example of Kafka use + Tutorials on understanding kafka better

• Piyush Garg/Kafka : Youtube Link ( No need to watch the implementation for now as it requires knowledge of backend in js,but the idea he tells is 🔥

[zakhaev26]

@punitkr03 bhaisahab aap ML me kab ghus gaye?

[punitkr03]

@zakhaev26 Suffering from skill issue.

[majorbruteforce]

So the idea is that we have seen the win probability of google during the IPL and international matches. How about we create a same system but its for our own college and GC!

This would be great in my opinion. Let's get the MVP rolling!

[majorbruteforce]

@zakhaev26 are we going to open a new repo for ML development or will it be under gc-server?

[geekofycoder]

@punitkr03 Suffering from Success syndrome on its way 😁

[zakhaev26]

@zakhaev26 are we going to open a new repo for ML development or will it be under gc-server?

It would be a good separation of concern if we create a new repository for it,but it shouldn't be just another dead part...I want ML guys to contribute there.Either way works...here also we can work on different branch..up to you guys

[geekofycoder]

@zakhaev26 let's discuss here and let me understand contributions part from you then with other members I will say what to do.

[geekofycoder]

[REF]https://docs.aws.amazon.com/msk/latest/developerguide/mkc-create-topic.html creating ec2 instance and installing kafka on ec2 machine

[uraharaSky]

What about the datasets....like what kind of data sets are we looking at?

[geekofycoder]

The dataset as I mentioned will be initially a dummy dataset but we will be creating custom dataset by deriving it from the dummy ones..... it's gonna take a lot of time so it is kept at last phase . Initially the model that is present on internet is manual one i.e. we have to feed the remaining balls reqd run rate etc....but we want it to be a real time dynamic application