Updates to Time Series Analysis in SQL Server

A couple months ago, I blogged about how my DataCamp course entitled Time Series Analysis in SQL Server soft launched. Since then, I’ve been hard at work, squashing bugs and making instructions and hints clearer and getting this course from soft launch to hard launch.

I have a new batch of updates coming out soon, so I wanted to walk you through some of the process. First, the average rating over the past eight weeks. The course launched about 10 weeks ago, so I’m stripping out that early period of first-movers who love everything and give it a 5-star rating.

It’s like watching your investment portfolio change from day to day.

The first batch of updates hit on July 16th, when I had enough data to learn where some of the problems were and fix them. That bumped the score up a bit, where it has mostly leveled off in the 4.4-4.45 range. Since then, I’ve had a few minor changes but nothing as big as the June 16th release or the upcoming release. My hope is that this next set of changes bumps me into the 4.5-4.6 range.

To figure out what to change, I get a helpful, sortable table with information on how people are doing. Here it is sorted by percentage of people who gave up and asked for the solution:

Psst, what’s the answer to question 8?

When 80% of the people are unable to answer your question, there’s a problem… Drilling in, we get to see the error messages learners get when trying out this problem:

I don’t think I correctly specified the 10 parameters, no.

From there, I can view the actual submitted code and see what my solution looks like versus the submitted code. Here’s the top-ranked example (with a bonus that you get the solution. If you found this blog post and remember it when going through my course, I think that’s worth one free solution):

Where it all went wrong.

With this information, I can modify the course. In this particular exercise, I decided to do two things: first, I provided the number of seconds and milliseconds to each DATETIMEOFFSETFROMPARTS() call. That was a common mistake where people filled in 8.0 for seconds.milliseconds rather than 7.999. Second, in the hint, I provide the exact date in a format which just happens to match the order of inputs. My hope is that these two things are enough to raise the success rate. For this example above, people put in 2039 as the year instead of 2038. Reading the improved hint, they’d see that the year is definitely 2038.

I ended up making these changes on about 15-20 exercises and will keep making changes to make the course successful. If you are going through the course and have direct feedback, hit the (!) button at the top-right of the screen on any exercise and leave your thoughts.

If you want to feel old, there are people old enough legally to drink in the United States who had not been born yet when this episode of the Simpsons first aired.

If you want to try this course out and learn all about working with times, dates, and time-related aggregations in SQL Server, get to the course on DataCamp. The first chapter is free, though subsequent chapters come at a cost. Just money, though, not souls or firstborn children.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s