jef-rey / data-sci

385- Intro to data science
0 stars 0 forks source link

p02 peer review summary #6

Open jmousavi opened 4 years ago

jmousavi commented 4 years ago

Data Preparation: You did a great job cleaning the new data, as well as gathering your old data. You also clearly explained what you accomplished at each step and it's easy to follow. One thing you could have cleaned better is the dates (as it gave you future problems).

Modeling: I liked how you explained what you would have done, had your model worked. However, as you were having issues merging your dates, I would suggest taking the temperature weather set, and getting an average temperature for each month & year. From there I would merge by month and year (you can potentially separate them out in your data set). There are also ways to get the month or year. For example: format(as.Date(date),"%Y")==2017). Since you did not get your model to work, you did not get a chance to communicate your results.

R proficiency: You did a good job demonstrating your R proficiency. You separated the code blocks into easy to understand chunks and paired it with a narrative. You also utilized the specialties within R, such as tapply. Since everything is organized and separated, it is easy to reproduce the code you wrote.

Communication: Great job on the communication! I enjoyed the narrative you had, it made it much easier to follow what you were doing at every step of the way. I loved your visualizations! They supplemented the new data you gathered very well. I also appreciated that you talked about what you would have done with your model, had your data merged.

Critical Thinking: You did a good job explaining why you were bringing in the second data set, and how it would supplement further insights to your old data set. However, I think you can have more discussion about what questions you are trying to answer or why these questions are important. Unfortunately you weren't able finish your model and provide deeper insights into your data set.

Feedback: For p03, you should figure out how to merge the data in order to show a model and see if any correlations exist. I would suggest creating a different kind of model (other then a linear model). Perhaps you can bring in another data set, other the temperature that can give more insights/correlations with your crime data set. Or you can potentially separate incidents with similar attributes (temperature/offense type/other variables), and group them together to see if any trends exist there. Since there is already a correlation between sunlight and depression, maybe you can tie in how many hours of sunlight there are.