Context This is the South African Lottery results from year 2000 when it started to 2015. I was interested in predicting whether there will be winners or not given the following publicly available information prior to betting: 1. Prize Payable 2. Rollover 3. Rollover Count 4. Next Estimated Jackpot The above mentioned features attract quite a lot of consumers and with an increase in the betters increase the chances of winning. This classifier is able to achieve 98% score and correctly predict against the X_test set on whether there will be a division 1 jackpot winner or not. Winner is 1 and no-winner is 0. The reason its 98% prediction is only because if there are 2 winners on division 1, it cannot predict this and hence if compared to the test set, it's not wholly accurate. Content The data was acquired from the National Lottery website. Please look at: https://www.nationallottery.co.za/lotto-history/?game=Lotto for further information Acknowledgements I am only new to machine learning, being a Chemical Engineer by vocation, I came across this sphere of knowledge and I must admit, most of my nights are spent just coding away and trying to predict the most ludicrous datasets I can dream up. However, its all been a lot of fun, and with every exercise I tend to learn a lot more. Inspiration One of my challenges is in visualising this data. I tried meshgrid and contourf plots, but getting errors. Also is it possible to to predict the number of division 1 winners? In the y_train data, there are a number of instances where there was more than 1 division 1 winners. However, the SVM was made only to be able to predict 0 for no winners or 1 for winners.