1. Conclusion - Where to go from here?
Congratulations on completing Modeling with Data in the Tidyverse. In this course we leveraged the data science toolbox you developed in previous courses to perform exploratory data analysis, fit both explanatory and predictive models, and study methods for model assessment and selection.
But where can we go from here?
2. R source code for all videos
First, a great way to learn how to code in a new language is to find code that you know works, then copy it, paste it, and tweak it to serve your goals. To help facilitate your new R modeling skills using the tidyverse, I've included all the R source code used in the videos for this course on GitHub. You can access this code using the URL above.
3. Other Tidyverse courses
Second, here is a link to other courses on DataCamp that use tidyverse. Being an effective data scientist requires you to develop a wide array of tools for your data science toolbox, and a lot of practice, practice, practice. These courses will help in this journey.
4. Refresher: General modeling framework
Perhaps the theory behind modeling interests you more. For example, recall our general modeling framework, which at its heart has a function f making explicit the relationship between y and x.
We kept things simple and only studied models f where f was linear. But by no means is one restricted to such models! What do we mean?
5. Parallel slopes model
Recall our parallel slopes model for house price as a function of size and condition. But why restrict ourselves to parallel lines?
6. Polynomial model
Here we have something known as a polynomial model where we allow for curvature by incorporating log10-size squared as an explanatory/predictor variable! This gives our model more flexibility.
Furthermore we're not restricted to models based on lines either! Yet another form of model are trees!
7. Tree models
Tree models are a form of triage. You start at the top of the tree, and based on answers to true/false questions, you progress down branches of the tree, where if the answer is true, you go left, if the answer is false, you go right. For example, say a house has log10_size = 3.2. Since 3.2 is < 3.387, you first go down the left branch. Next, since 3.2 is !< than 3.184, you then go down the right branch. This model's fitted/predicted value of this house's log10_price is 5.642, or about $438K. Repeating this triage for all 21k houses, there are 8875 houses that fall into this final branch.
8. DataCamp courses using other models
We've only scratched the surface of possible other models to consider. Here are other DataCamp courses you can take that consider more complex, but also potentially, more powerful models.
9. Refresher: Regression table
Finally, recall the regression table of the model of teaching score as a function of age. We only looked at values in the estimate column, like the negative slope for age of -0.006, suggesting that as professors age, they tend to have lower teaching scores. But what do the other columns tell us?
They speak to the "statistical significance" of our results. For example, can we conclusively say that age and score are negatively related for ALL instructors, or was this relationship just a fluke occurrence for these 463 instructors? How would these results differ if we selected 463 different instructors? To be able to answer questions like these, we need to understand statistical inference.
10. ModernDive: Online textbook
If you are interested in statistical inference, we suggest you check out moderndive, an electronic textbook that Chester Ismay and I co-authored.
-moderndive uses the same tidyverse tools as in this course
-expands on the regression models from this course and others
-uses the evals and house_prices datasets and others
-all towards the goal of learning statistical inference via data science.
-It's available at moderndive.com
11. Good luck!
I hope you had fun in this course, and continue to enjoy your data science journey! A special thanks goes to any student who's ever taken a class with me; I couldn't have created this course without you.