Python and Jupyter Intro

Machine Learning and SKLearn

ArcGIS and Python API

AGOL Applications

Suave with SDGs and Spatial Stats

Comments about modules#hidden

Additional lectures#hidden

HW1 - Python and Jupyter Intro

HW2 - Geopandas and Spatial Analysis

HW3 - ML and Geoenrichment, ArcGIS

HW4 - Deep Learning

HW Comments#hidden

Emphasis on ML or Spatial Analysis?

Discussions during lectures?

Hours spent per week#number

Most valuable component#hiddencopy

Fit GPS curriculum?#hidden

Other comments#hidden

Machine learning - prior proficiency

Deep learning - prior proficiency

Python - prior proficiency

GIS - prior proficiency

Geopandas - prior proficiency

ArcGIS Desktop or AGOL - prior proficiency

ArcGIS Python - prior proficiency

Spatial stats - prior proficiency

Raster analysis - prior proficiency

It is so hard to suggest reducing any particular topic - each deserves a course of it's own! However, I think the most valuable topics we cover (what students will use beyond this class) are (a) using Python & creating projects in Jupyter notebooks, (b) geopandas, and (c) sklearn. However, the other topics are all so interesting that I don't want to suggest reducing them. I wish this course could be broken into 2 quarters: the first section would focus essentially on the topics covered in the first two homeworks, and the second section would cover special topics and give students a chance to develop independent projects.

More spatial statistics, but I'm biased.

I think that it isn't so much a question of accompanying modules with homework as it is clarity of the homework questions. I think HW2 was challenging because it was hard to tell (as a student) if the misunderstandings were caused because of a lack of knowledge in Python OR if the question was vague. In retrospect, I think some of the questions were a bit unclear. However, I want to thank you for these homework assignments - they really helped me personally grow and learn. I appreciate the effort that went into designing the homework and I found the exercises challenging in a good way.

More spatial data in Python

I am definitely going to use Jupyter notebooks for formatting projects that I'm working on. I love how interactive the notebooks are and I hope to see more social scientists using them. I hope that this can help in the replicability crisis. I am also thinking of future projects that will likely use sklearn and geopandas.

I have only taken advanced data courses applying econometrics to datasets, but one of those courses was similar to this. The econometrics course I took was different in that we would look at models mathematically and then apply them in R. This might help students understand what's going on "behind the scenes" with some concepts.

Thank you for teaching this class! I feel very happy to have had the opportunity to take this class and work on these cool topics!

Limited exposure

Limited exposure

Limited exposure

Limited exposure

More machine learning

Limited exposure

Limited exposure

Limited exposure

The modules that are for expanding are those that I believe have the most cross cutting work, and which we might be able to continue to play with after we are done with the course. Using ArcGIS might be the standard, but it is also something that is proprietary, and so restricted.
The python and pandas are extremely useful no matter where any of us end up, but the ArcGIS requires someone to be paying us to really use. They were also the part where we had to have the best grounding before moving forward into any of the other areas.
Deep Learning section was something where I just followed the directions, but understood nothing of what I was doing.
The rasters are useful to know because they most directly built on what we have done before, but we went so fast that even with that grounding I was still lost.

The first homework was easy because most of the problems were so easily google-able, while the HW2 was very much not. So the switch was problematic for us to get through. The shift kind of knocked us off the learning curve, and left most of us behind.

More machine learning

The pandas data manipulation and general rules on accessing APIs complement skills we already have on data analysis quite well. I had previously been forced to bungle my way through both pandas and apis to get data earlier, and this course's early material will make subsequent work much more ordered and easy to read.

The curriculum is quite an outlier, but not when compared to the previous GIS course. Both of these focus heavily on skills for the sake of learning them. Most GPS courses focus heavily on policy specific applications, and the skills are learned so that they can be applied in a specific area.

Limited exposure

More machine learning

Limited exposure

Limited exposure

More spatial data in Python

Limited exposure

Limited exposure

More spatial data in Python

Jupyter Notebooks, Pandas and Geopandas.

The course provided us with great jupyter notebooks, but I wish we has more time to go through them in class. I feel like we skipped a few initial steps (that were taken as given) that were important for our general understanding.

Appreciate the Professor and TA's patience with our limited knowledge of Python.

Limited exposure

More machine learning

machine learning

it is an outlier a bit as it uses python, but it is as useful it not more useful than other courses at GPS.

I believe having mini python exercises one a week to get the students familiar with code and working in jupyter notebook will be very helpful

Limited exposure

Can reduce the amount of work to be done but a better sense of what and why we are doing. like the machine learning assignment was pretty well laid out in terms of getting what we were doing and why

More spatial data in Python

ArcGIS online, network analysis (location - allocation), geoenrichment

Limited exposure

Limited exposure

Limited exposure

Limited exposure

Limited exposure

More spatial data in Python

Using ArcGIS to build complex StoryMaps and Dashboards for presenting geospatial data in a visually-pleasing and easy to understand way to stakeholders and partners. Using geopandas to sort, analyze, and prepare geospatial data for visualization as I thought it was much easier than using QGIS/R.

It does fit into the curriculum, but I wish we had more work on creating and presenting geospatial data to an audience. Our final projects do that, but I wish our homeworks had us practice more (such as a homework based on Amy's guest lecture).

Limited exposure

We work with rasters a lot and should probably deal with them earlier. Deep learning was way too hard (and probably not relevant)

All of them were fine except deep learning

More spatial data in Python

Geopandas, loops

Outlier, but in a good way. I appreciated that it was all about making and analyzing maps, and that we didn't have to write busywork essays and could focus just on the python. However, the last 4 weeks were pretty hard and maybe not applicable to studying policy

For next time, it might be helpful to think of GPS students as computer-illiterate.
This class was fun, thank you

Limited exposure

The first assignment was interesting because the first two problems seemed disconnected from the rest of the assignment, and felt disproportionately more difficult than the rest of the homework which was very easy because it was the same as the examples with just a slight change. The second homework was very confusing for me and I didn't feel like I fully understood what I was doing, the third assignment was an ideal difficulty. Conceptually I understood what I was doing on the third assignment, and the code was very similar to what we had gone over in lecture. The last deep learning homework was difficult just because I generally understood what was happening conceptually but it felt more like a lab where I was blindly following directions and didn't necessarily always understand why I was doing every step.

More machine learning

I think the exposure to jupyter notebook and python as well as machine learning are components that I found riveting and hope to use in the coming years.

Limited exposure

Limited exposure

There was a big jump in difficulty from HW1 and HW2. It would be helpful to have the homework structured in a way that someone can continue if they get stuck on a certain step (similar to HW4).

The balance is about right

Conceptually, the course fits into the curriculum. There is just a learning curve from learning a new language. It is one of the more quant heavy courses.

I think it may be interesting to discuss some of the readings in class. Some mini quizzes for participation could be helpful as a concept check.

Limited exposure

Limited exposure

Limited exposure

Limited exposure

I really appreciate the notebooks being available in order to review the code again after lecture.

Some of the homework assignments were challenging but I think I was able to complete the majority of them correctly and after some additional help in office hours. I think homework two could have been two assignments.

More spatial data in Python

I feel like I have a better understanding of python and I have a better understanding of using python and GIS applicatons.

It is similar by using GIS applications to test a hypothesis but different in that we used python and expanded our data science skills. I think this fits into the curriculum. I would like to have seen more international examples which would link to other GPS courses as well.

Limited exposure

Limited exposure

Limited exposure

I think spending a bit more time on the introduction to coding in Python would be really valuable, as most of us did not have much prior background in coding. I also think spending more time on the machine learning and especially the deep learning would be valuable (especially the concepts and theoretical understanding behind it). It may be possible to reduce the section about AGOL applications or guest lectures on other topics in order to spend more time learning these methods.

The modules incorporated in the homework made sense. However, it felt as though there was a steep learning curve from Homework 1 to Homework 2. Spending more time on Python basics (examples of for loops, etc) may have helped to then move on to Homework 2. Homework 3 was conceptually reasonable; however, it was very time consuming as we applied similar machine learning techniques on different datasets. Perhaps some of this repetition could be removed to reduce the time spent on this homework. I also wonder if providing code on how to conduct deep learning in Python as a resource would be helpful in addition to Homework 4 that was primarily in ArcGIS.

More machine learning

I will definitely continue using Python for spatial analyses, and I will continue to build my coding competency in order to not rely on ArcMap or ArcGIS Pro. I will also continue to learn how to utilize machine learning.

Really appreciated this class! I've been wanting to learn Python and machine learning concepts for a long time, and this class was great. I will definitely continue expanding my coding abilities in Python and continue learning about machine learning. Thanks so much!

Limited exposure

Limited exposure

Limited exposure

More spatial data in Python

Limited exposure

Limited exposure

Limited exposure

The site that hosted the data used in HW 3 kept crashing and made finishing homework in a timely manner difficult.

More spatial data in Python

having an introduction to python

This course was extremely challenging. This course is teaching us a separate coding language a large majority of us are unfamiliar with, and it is difficult to keep up with our other demanding class schedules. I think this class is interesting and teaches valuable skills, i just think it should be either worth more units, or divided into two courses.

Limited exposure

Limited exposure

Limited exposure