1. Final thoughts
OK, we're at the finish line! Congratulations on going through all the lessons!
2. What we've learned
Let's quickly recap what we have learned in this course.
First of all, we've learned what Kaggle actually is and the Machine Learning competitions process.
Then, we've covered all the steps we need to perform in any competition:
starting with a problem definition
and some initial data exploration,
develop a reliable validation strategy,
test hypothesis, generate new features,
and finally, build model ensembles.
These topics have been covered on relatively accessible examples. Now it's your turn to start competing on Kaggle and expand your knowledge and experience.
3. Kaggle vs Data Science
The final note I'd like to emphasize is that Kaggle competitions are not equal to Data Science. They cover only some subareas of the real Data Science job.
4. Kaggle vs Data Science
First of all Data Science is not only about building the Machine Learning models. Tasks could also include some Data Analytics stuff in order to provide insights, make analytical reports and help decision makers. Kaggle does not help here.
5. Kaggle vs Data Science
But even Machine Learning projects in Data Science job include some additional steps that are not covered by Kaggle.
Let's list the usual stages performed to develop the model.
We need to start with the business people communication to get the model requirements.
Then we collect data needed for the model. Either it is located in .csv files, in databases or Hadoop cluster.
Based on the problem, we have to choose the metric to be optimized.
And make a fair train/test split for the model evaluation without a leakage.
In the competitions, Kaggle does all these steps above for us.
What we do on Kaggle is creating the best performing model itself. Of course, it is a long process and includes all the stuff you've seen in this course.
Finally, in the real project, we need to incorporate model created into the production environment. Here Kaggle also does not help.
6. Insert title here...
However, Kaggle does give practical skills and teaches tricks that are not covered in any online courses or books. One could get hands-on experience for any problem type within the community of the best Data Scientists in the world.
7. Start competing on Kaggle!
Congratulations again on finishing this course, and thank you for taking it! I hope you've enjoyed this course, as much as I've enjoyed teaching it.
Now go straight to kaggle.com and take part in Machine Learning competitions! Good luck!