Adjustments of the LDSSA structure to the online form

[majkah0]

In this discussion, I'd like to draw attention to a problem that the Academy is facing in recent years and propose some solutions that stem from discussions within the taskforce group. I would like to hear your opinions on both the problem and the proposed solutions. This discussion has a lot of points, please provide opinions about as many as you can.

Deadline: 25. 9. 2023

Adjustment of the LDSSA structure to the online form and measures to improve student success

The current form of the bootcamp was designed for an in-person bootcamp weekend. Students and instructors met for two consecutive days with presentations and exercise solving. A 15 min presentation followed by 45 min exercise solving was reserved for each SLU. The students could obtain help from the instructors in case of difficulties. The students could then review the material and finish any remaining exercises during two additional bootcamp weeks.
This structure was preserved when the academy moved to the remote form due to the pandemic in batch4. The bootcamp weekend presentations were done online and to account for the less intensive instructor support, the bootcamp was extended to three weeks in batch5 and batch6.
The graduation rate from the academy is 35 - 60%. The attrition is progressive, with the largest share in S01 (10 - 25%) and the capstone (20 - 30%). For statistics, see here https://docs.google.com/spreadsheets/d/18ShmxDAcJi3X3iuFy_lZYxmjj-4Gc0KmdUWkH2CO2H0/edit#gid=0 . The data for batch 2 and 3 is incomplete and only approximate.
The following proposals aim to address these problems.

1. No exercise solving during the bootcamp weekend

Problem: Students are not solving the exercises during the allocated time in the bootcamp weekend because they are not able to grasp the topic so fast and there is no instructor support to compensate for it. This time is basically wasted and it unnecessarily extends the bootcamp weekend.

Implementation: The solution that we are planning to test in batch7 is to have just the presentations, possibly slightly longer, and omit the exercise solving.

2. Split the bootcamp weekend into two parts

Problem: The material presented in the boot camp weekend is too much to grasp for such a short time. The later SLUs require a good understanding and digesting of the previous SLUs.

Implementation: The solution that we are planning to test in batch7 is to split the bootcamp weekend in two parts, SLU04 - SLU10 and SLU11 - SLU17 (possibly also SLU18 and 19 here). Each part would be presented on one Sunday morning on different weekends. The duration of each session would be about 4 hours.

3. Release the learning notebooks before the bootcamp

Problem: Students are not able to optimally use the time reserved for questions on the bootcamp weekend. They would be better prepared and come with deeper questions if they could read the learning notebooks beforehand.

Implementation: The solution that we are planning to test in batch7 is to release the learning notebooks one week before the bootcamp presentations - at the beginning of the setup week for SLU04-10 and at the beginning of the first bootcamp week for SLU11-19.

4. Create training hackathons

Problem: According to student feedback, most students arrive at the first hackathon feeling very unprepared and having a hard time going through the workflow. This can result in an unpleasant and frustrating experience and also in a suboptimal learning outcome. The gap between the exercise notebooks and the hackathon is too large.

Implementation: The solution that we are planning to test in batch7 is to create two optional guided training hackathons in the form of an exercise notebook. The first training hackathon should go after SLU10 when students know how to perform data cleaning, EDA, and use two models. The second training hackathon should be an extended form of SLU16, the workflow SLU.

5. Extend the bootcamp by one week to four weeks

Problem: The bootcamp contains a large amount of material and many new concepts compared to the other specializations. This material is also the basis for the rest of the academy. The students need to master skills that will be used over and over during the whole batch, in every hackathon and in the capstone project. At the same time, the allocated time is one week shorter than in the other specializations and the students are beginners. This reflects in the student experience - students report too much work and not enough time to properly digest the material. The measurement of the time spent on the SLU completion reflects this (although the data is limited). There is also a large amount of attrition which is problematic given that the places are limited and also our effort spent in the admission process.

Implementation: The solution would be to extend the bootcamp by one week to a duration of four weeks, same as in the other specializations. The four week deadline also allows for some slack in case of life getting in the way. Together with the previous suggestions, it could lead to higher student success during the whole academy and to a better student experience. In batch7, we will keep a three week bootcamp. The consensus in the teaching AOR is that three weeks are enough to go through the material and solve the exercise notebooks. Nevertheless, it would be good to hear more opinions. The question also remains if three weeks are enough to thoroughly digest the material and get a good basis for the rest of the academy, as opposed to just completing the assignments on time.

[Gustavo-SF]

1. No exercise solving during the bootcamp weekend

Problem: Students are not solving the exercises during the allocated time in the bootcamp weekend because they are not able to grasp the topic so fast and there is no instructor support to compensate for it. This time is basically wasted and it unnecessarily extends the bootcamp weekend.

Implementation: The solution that we are planning to test in batch7 is to have just the presentations, possibly slightly longer, and omit the exercise solving.

If the 4 hours you propose include the extension of the presentations, then I think it is completely ok. I would still leave it up to the instructor how much time they want to use.

[Gustavo-SF]

Together with this, I would have instructors available some other time to take any questions from the exercises.

[Gustavo-SF]

2. Split the bootcamp weekend into two parts

Problem: The material presented in the boot camp weekend is too much to grasp for such a short time. The later SLUs require a good understanding and digesting of the previous SLUs.

Implementation: The solution that we are planning to test in batch7 is to split the bootcamp weekend in two parts, SLU04 - SLU10 and SLU11 - SLU17 (possibly also SLU18 and 19 here). Each part would be presented on one Sunday morning on different weekends. The duration of each session would be about 4 hours.

This has been proposed for some time, and with the remote format I think it is the best option too. It is veery hard to absorb so much information through a video call.

[Gustavo-SF]

3. Release the learning notebooks before the bootcamp

Problem: Students are not able to optimally use the time reserved for questions on the bootcamp weekend. They would be better prepared and come with deeper questions if they could read the learning notebooks beforehand.

Implementation: The solution that we are planning to test in batch7 is to release the learning notebooks one week before the bootcamp presentations - at the beginning of the setup week for SLU04-10 and at the beginning of the first bootcamp week for SLU11-19.

The only thing I see about this is that we have to have the materials ready by then. Do we have enough time? If yes, I think this is an easy thing to implement and we can go with this.

[Gustavo-SF]

4. Create training hackathons

Problem: According to student feedback, most students arrive at the first hackathon feeling very unprepared and having a hard time going through the workflow. This can result in an unpleasant and frustrating experience and also in a suboptimal learning outcome. The gap between the exercise notebooks and the hackathon is too large.

Implementation: The solution that we are planning to test in batch7 is to create two optional guided training hackathons in the form of an exercise notebook. The first training hackathon should go after SLU10 when students know how to perform data cleaning, EDA, and use two models. The second training hackathon should be an extended form of SLU16, the workflow SLU.

This is a high-effort implementation and might be difficult to ensure this. I am hesitant about this one.

[Gustavo-SF]

5. Extend the bootcamp by one week to four weeks

Problem: The bootcamp contains a large amount of material and many new concepts compared to the other specializations. This material is also the basis for the rest of the academy. The students need to master skills that will be used over and over during the whole batch, in every hackathon and in the capstone project. At the same time, the allocated time is one week shorter than in the other specializations and the students are beginners. This reflects in the student experience - students report too much work and not enough time to properly digest the material. The measurement of the time spent on the SLU completion reflects this (although the data is limited). There is also a large amount of attrition which is problematic given that the places are limited and also our effort spent in the admission process.

Implementation: The solution would be to extend the bootcamp by one week to a duration of four weeks, same as in the other specializations. The four week deadline also allows for some slack in case of life getting in the way. Together with the previous suggestions, it could lead to higher student success during the whole academy and to a better student experience. In batch7, we will keep a three week bootcamp. The consensus in the teaching AOR is that three weeks are enough to go through the material and solve the exercise notebooks. Nevertheless, it would be good to hear more opinions. The question also remains if three weeks are enough to thoroughly digest the material and get a good basis for the rest of the academy, as opposed to just completing the assignments on time.

I still think three weeks is enough, but since I don't have a strong opinion about keeping it shorter, I would accept 4 weeks in the following scenarios:

• Makes our calendar a better fit for the students because the Bootcamp will overlap with Christmas
• Makes our calendar a better fit for the instructors since the date for the Hackathon and BLU delivery works better for everyone.