Evaluation notebooks break when run

[singhish]

This is a recurring issue for each evaluation notebook. When the line

tradeoff_df["quality"] = tradeoff_df.role.apply(lambda r: r2q_map[r])

is run, execution fails with the following error:

KeyError                                  Traceback (most recent call last)
<ipython-input-23-21c779f8c074> in <module>
      1 # Make a number so that can get the plots to come out in order
----> 2 tradeoff_df["quality"] = tradeoff_df.role.apply(lambda r: r2q_map[r])
      3 tradeoff_df["count_diff"] = tradeoff_df[["count"]] - 1

~/miniconda3/envs/emissioneval/lib/python3.8/site-packages/pandas/core/series.py in apply(self, func, convert_dtype, args, **kwds)
   3846             else:
   3847                 values = self.astype(object).values
-> 3848                 mapped = lib.map_infer(values, f, convert=convert_dtype)
   3849 
   3850         if len(mapped) and isinstance(mapped[0], Series):

pandas/_libs/lib.pyx in pandas._libs.lib.map_infer()

<ipython-input-23-21c779f8c074> in <lambda>(r)
      1 # Make a number so that can get the plots to come out in order
----> 2 tradeoff_df["quality"] = tradeoff_df.role.apply(lambda r: r2q_map[r])
      3 tradeoff_df["count_diff"] = tradeoff_df[["count"]] - 1

KeyError: 'MAMFDC'

The error is the exact same across all evaluation notebooks except for Evaluate_power_vs_trip_start_end_analysed_master.

[shankari]

As we discussed in person, the most recent version of the notebooks was lost when I ran git checkout HEAD in the root directory by mistake. Fortunately, I do have the checkpoint files, and some files that were never actually checked in, to provide a backup. For each of the evaluation files, we need to compare the main file with the checkpointed file and resolve the differences.

More details in #12

[shankari]

Grepped through all notebooks and the r2qmap is consistently the same.

$ grep r2q_map new_versions/*.ipynb | grep MAMFDC | wc -l
       0

I guess there might have been a version in which I added MAMFDC, but if so, it is lost to the bad checkout.
It could also be that I collected MAMFDC results primarily after the notebooks were complete.

Ah, checking my thesis, I don't have any MAMFDC results, so I presumably collected that data only after the bulk of the analysis was complete. So the fix is trivial.

[shankari]

The fix for this is

-    "r2q_map = {\"power_control\": 0, \"HAMFDC\": 1, \"MAHFDC\": 2, \"HAHFDC\": 3, \"accuracy_control\": 4}\n",
-    "q2r_map = {0: \"power\", 1: \"HAMFDC\", 2: \"MAHFDC\", 3: \"HAHFDC\", 4: \"accuracy\"}"
+    "r2q_map = {\"power_control\": 0, \"HAMFDC\": 1, \"MAHFDC\": 2, \"HAHFDC\": 3, \"MAMFDC\": 4, \"accuracy_control\": 5}\n",
+    "q2r_map = {0: \"power\", 1: \"HAMFDC\", 2: \"MAHFDC\", 3: \"HAHFDC\", 4: \"MAMFDC\", 5: \"accuracy\"}"

We should discuss the timeline for copying over the checkpoints and verifying the results. I can copy over the checkpoints first, but I am not sure I will have time to verify each of the results against the thesis.

[shankari]

I also needed to make the graphs bigger - e.g.

-    "ifig, ax_array = plt.subplots(nrows=2,ncols=3,figsize=(9,6), sharex=False, sharey=True)\n",
+    "ifig, ax_array = plt.subplots(nrows=2,ncols=3,figsize=(16,6), sharex=False, sharey=True)\n",

to accommodate the new MAMFDC entry

-    "ifig, ax_array = plt.subplots(nrows=4,ncols=3,figsize=(10,10), sharex=False, sharey=True)\n",
+    "ifig, ax_array = plt.subplots(nrows=4,ncols=3,figsize=(16,10), sharex=False, sharey=True)\n",

But that is a purely cosmetic fix. The notebook will run through without any errors even without these fixes.