Add exercises

README.md

@@ -32,16 +32,18 @@ The slides for this workshop can be viewed on the [website of this course](https
 
 ### Exercises
 
-TODO
+The exercises for this course are listed in the quarto markdown files which can be found in the [exercises](exercises/) directory. The tasks are
+divided into those that can be answered/solved during the session, and also "Extension" tasks which can be worked on outside of the session.
 
-The exercises for the course can be found in the [exercises](exercises/) directory.  
-These take the form of \<partially completed jupyter notebooks/downloadable code/online tasks or games etc.\>.
+There is a small group of python scripts in the [problem](exercises/problem/) sub-directory. These act as an example of a non-reproducible workflow. The tasks involve understanding and 
+improving the reproducibility of the scripts using what is covered in the session. It is recommended to download these scripts to a seperate location on your local machine for 
+this purpose.
 
 ### Worked Solution
-TODO
 
 Worked solutions for all of the exercises can be found in the [worked solutions](worked-solutions/) directory.  
-These are for recapping after the course in case you missed anything, and contain example solutions.
+These cover the tasks that require any modifications/improvements to the example workflow provided. They are not exhaustive, but use common technologies and approaches
+to solving the tasks.
 
 ### Mini ReproHack
 
@@ -66,7 +68,11 @@ If you want to follow along with the exercises, basic Python coding skills and a
 
 ## Installation and setup
 
-TODO
+To complete the exercises, you will need to download the dataset used by the python scripts. This can be found [here](https://www.metoffice.gov.uk/hadobs/hadcrut5/data/HadCRUT.5.0.0.0/download.html)
+As part of the exercises will be assessing the data dependencies of the scripts, it will be left as a task to find out which of the datasets you need!
+
+It is recommended to download the python scripts from [exercises/problem](exercises/problem/) to your local machine, in a new directory.
+This will allow you to develop the code as you would in practice.
 
 
 ## License

exercises/README.md

@@ -0,0 +1,20 @@
+# Exercises in reproducibility
+
+After going through the slides on how to make our research projects reproducible,
+it is time to go through a practice scenario!
+
+The code in `problem/` is an example of a research project which has not taken 
+reproducibility into account.
+
+## Data
+
+If you are trying to work through these exercises, you will need the data that
+the scripts rely on, which can be found here: https://www.metoffice.gov.uk/hadobs/hadcrut5/data/HadCRUT.5.0.0.0/download.html.
+
+## Instructions:
+
+1. Create a new repository on your local machine with `git init repro-examples`
+2. Copy the files in the `problem` directory into the new repository
+3. Download the data needed for the analysis from https://doi.org/10.5281/zenodo.7014332
+4. Try to complete the tasks in the files named `*_exercises.qmd`. Some of these are
+achievable during the session, others are tasks that could be worked on in your own time.

exercises/automation_exercises.qmd

@@ -0,0 +1,9 @@
+## Automation Exercises
+
+1. What tools/technologies could be used to automate producing the plots?
+2. How could we automatically fetch the data needed for the plots
+
+### Extension
+
+1. Create an automated pipeline for producing all the figures
+2. Automate the fetching of the necessary data

exercises/dependencies_exercises.qmd

@@ -0,0 +1,14 @@
+## Dependencies exercises
+
+Returning to the Python scripts, try to answer the following questions:
+
+1. What are the python dependencies? 
+2. What are the system dependencies?
+3. What are the data dependencies?
+
+### Extension
+
+Using your favourite dependency management tool (e.g. Conda, Nix), create a
+reproducible environment configuration to handle the python and system dependencies.
+
+Create a way to automatically fetch the data dependencies when using the scripts.

exercises/documentation_exercises.qmd

@@ -0,0 +1,12 @@
+## Documentation Exercises
+
+Tasks:
+
+1. Create a README for the repository
+2. Add comments and docstrings where you think is necessary
+
+### Extension
+
+Use an documentation tool, such as Sphinx, to automate building the
+documentation.
+

exercises/final_exercises.qmd

@@ -0,0 +1,15 @@
+## Try it out!
+
+Tasks:
+
+1. Try running some of the worked examples in the repository
+  - Which ones work? Which do you find the easiest to use?
+
+### Extensions
+
+1. Share your repository with someone, see if they can reproduce the plotd
+2. Try to reproduce the results using the repo that you have been shared
+3. Try to use a different dataset (e.g. a different version of the HadCrut data)
+   - Does it work?
+   - What changes did you need to make?
+

exercises/problem/plot1.py

@@ -1,7 +1,24 @@
 import xarray
 import matplotlib.pyplot as plt
 
-ds = xarray.open_dataset("../data/CARS2009_temp20-40S_69-88W.nc")
+# Recreate the "warming stripes" visualisation
+# https://showyourstripes.info/
+
+ds = xarray.open_dataset("../data/HadCRUT.5.0.0.0_analysis_summary-series_global_annual.nc")
+
+fig, axs = plt.subplots(2, 1, figsize=(10,6), layout='constrained')
+
+x = ds['time'].data
+
+y_lower = ds['tas_lower'].data
+y = ds['tas_mean'].data
+y_upper = ds['tas_upper'].data
+
+axs[0].pcolor(y.reshape(1, -1), cmap="coolwarm")
+axs[0].axis('off')
+
+axs[1].errorbar(x, y, yerr=[y - y_lower, y_upper - y])
+axs[1].set_ylabel("Annual Temperature Anomalies Compared to 1961-1990, \N{DEGREE SIGN}C")
+axs[1].set_xlabel("Year")
 
-ds['temp'].plot()
 plt.show()

exercises/problem/plot2.py

@@ -0,0 +1,28 @@
+import xarray
+import matplotlib.pyplot as plt
+
+global_ds = xarray.open_dataset("../data/HadCRUT.5.0.0.0_analysis_summary-series_global_annual.nc")
+northern_ds = xarray.open_dataset("../data/HadCRUT.5.0.0.0_analysis_summary-series_northern-hemisphere_annual.nc")
+southern_ds = xarray.open_dataset("../data/HadCRUT.5.0.0.0_analysis_summary-series_southern-hemisphere_annual.nc")
+
+def plot_temp_with_error(ax, ds, label):
+    axs.fill_between(ds['time'].data, ds['tas_lower'].data, ds['tas_upper'].data,
+                     color="grey", alpha=0.3, edgecolor=None)
+    axs.plot(ds['time'].data, ds['tas_mean'].data)
+
+
+fig, axs = plt.subplots(1, 1, figsize=(10,6), layout='constrained')
+
+plot_temp_with_error(axs, global_ds, label="global")
+plot_temp_with_error(axs, northern_ds, label="northern")
+plot_temp_with_error(axs, southern_ds, label="southern")
+
+t_min = global_ds['time'].min()
+t_max = global_ds['time'].max()
+
+axs.hlines(0.0, t_min, t_max, color="black", linestyle="--")
+
+axs.set_xlabel("Year")
+axs.set_ylabel("Annual Temperature Anomalies Compared to 1961-1990, \N{DEGREE SIGN}C")
+
+plt.show()

exercises/problem/plot3.py

@@ -0,0 +1,39 @@
+import xarray
+
+import matplotlib.pyplot as plt
+import matplotlib as mpl
+
+ds = xarray.open_dataset("../data/HadCRUT.5.0.0.0_analysis_summary-series_global_monthly.nc")
+
+fig, axs = plt.subplots(1, 1, figsize=(10,6), layout='constrained')
+
+N_years = ds.time.size / 12
+
+alpha = 0.001
+delta_alpha = (1.0 - alpha)/N_years
+
+year_group = ds.groupby("time.year")
+
+years = [ year for year, _ in year_group]
+
+first_year = years[0]
+last_year = years[-1]
+
+delta_years = last_year - first_year
+
+cmap = mpl.colormaps['rainbow']
+
+for year, year_data in year_group:
+    alpha += delta_alpha
+    yearly_mean = year_data['tas_mean'].mean()
+    axs.plot(year_data.time.dt.strftime("%b"), year_data['tas_mean'], 
+             label=year, color=cmap((year - first_year)/delta_years), alpha=alpha)
+
+
+axs.set_ylabel("Temperature Anomalies Compared to 1961-1990, \N{DEGREE SIGN}C")
+axs.set_xlabel("Month")    
+
+colorizer = mpl.colorizer.Colorizer(norm=mpl.colors.Normalize(first_year, last_year), cmap='rainbow')
+fig.colorbar(mpl.colorizer.ColorizingArtist(colorizer), ax=axs)
+
+plt.show()

exercises/version_control_exercises.qmd

@@ -0,0 +1,16 @@
+## Version Control Exercises
+
+Hopefully, at this point in the week, you will all be comfortable with version control!
+
+The exercises in this course will work on improving an example set of python scripts,
+to make them more reproducible!
+
+These scripts are hosted on the same GitHub repository as these slides:
+(https://github.com/Cambridge-ICCS/ReproducibilityInComputingCourse)
+Specifically in the directory `exercises/problem`.
+
+Tasks:
+
+1. Download the Python scripts from the GitHub repository
+2. Create a new git repository on your local machine
+3. Add the Python scripts to git for version control!

src/slides.qmd

@@ -15,14 +15,18 @@ authors:
 {{< include introduction.qmd >}}
 
 {{< include version_control.qmd >}}
+{{< include ../exercises/version_control_exercises.qmd >}}
 
 {{< include dependencies.qmd >}}
+{{< include ../exercises/dependencies_exercises.qmd >}}
 
 {{< include automation.qmd >}}
+{{< include ../exercises/automation_exercises.qmd >}}
 
 {{< include testing.qmd >}}
 
 {{< include documentation.qmd >}}
+{{< include ../exercises/documentation_exercises.qmd >}}
 
 {{< include fair_principles.qmd >}}