October 15, 2019 0

IPL – 1st Inning Total Prediction using Machine Learning Models

IPL – 1st Inning Total Prediction using Machine Learning Models


so we saw in the last video how you can
get the IPL data from kaggle and looking around there’s this excellent article by
the cricket servant URL in the description below the video where it
shows the kind of stuff that you can do once you have the data so here you know
you’ve got the various overs in a cricket match and you can kind of see
how the match is going the blue line is the predicted score by the model and the
red line is the actual score and you can see here that pretty rapidly it zooms in
so the model even in the first few overs has figured out that oh this match is
gonna have a target of our nearly 180 what if you just looked at the over
eight it’s very volatile it seems to go oh it’s I think it’s gonna be a hundred
seventy if it goes at this rate and then you have an over two with few runs and
it crashes down to oh I think it’s only gonna get a hundred forty seems up here
so it’s all over the place but using the machine learning model he’s getting a
far more accurate projection so he’s never off by more than really 10 15 runs
particularly if you wait till five or six overs which you know you’re supposed
to do with dockwood Luis so basically wait for five or six overs and the model
does pretty well he looks at a couple of models so he
does a multiple regression he does random forests and I think it
was as SVM Oh No nearest neighbor yeah nearest neighbor regressive you can’t
you know if you have a random forest model you can’t connect oh oh and the
numbers are these three numbers but basically the nearest neighbor and
random forests are the better models like with the less volatility but even
using this so he he gives the fear like you know what let it go there here’s the
formula the final final score is the current score less than one point one
six times the number of balls remaining minus four point zero four times the
number of wickets taken plus seventeen point one so this is like the Duckworth-Lweis calculator that we’ve been looking up before and you can give it a
go you can just plug in those numbers to a match and see how well it does you can
alter back test you know like the the back testing video we did the other day
I’ll put that link in in the description below the video as well so then you can
just look at the various IPL matches that have been played so far this year
look at this Oh after the third ball in the seventh over the score was this
there were this many balls left there were this many wickets taken
so this predicted that the score would be hardened eighty
and then you can check to see Oh in the end the score was 184 but you can
experiment with this model and see you know it is pretty good so this is just
the kind of thing that you can be thinking about so once you have the data
from coggle you can load it up into either python or or whatever system
you’re using and you can produce the various models that are available

Leave a Reply

Your email address will not be published. Required fields are marked *

© Copyright 2019. Tehai. All rights reserved. .