complete revision

.gitignore

@@ -43,3 +43,4 @@ paper/**/*.out
 paper/**/*.synctex.gz
 paper/**/*.blg
 paper/**/*.bbl
+texput.log

code/khan_helpers/khan_helpers/functions.py

@@ -170,7 +170,7 @@ def corr_mean(rs, axis=None, fix_inf=False, **kwargs):
         Axis or axes along which the means are computed. If `None`
         (default), the mean of the flattened array is computed.
     fix_inf : bool, optional
-        See `z2r()` docstring for details. Default: False.
+        See `r2z()` docstring for details. Default: False.
     **kwargs : various types, optional
         Additional keyword arguments passed to `numpy.nanmean` (see
         https://numpy.org/doc/stable/reference/generated/numpy.nanmean.html
@@ -622,7 +622,7 @@ def preprocess_text(textlist, correction_counter=None):
     POS tagging, and lemmatization.
 
     Occasionally, the Treebank POS tagger mis-tags a word, which causes
-    WordNet's "Morphy" to apply the morphologocal transformations and
+    WordNet's "Morphy" to apply the morphological transformations and
     detachment rules for the wrong syntactic category, and fail to
     lemmatize the word.  The function attempts to handle these
     instances and can optionally record corrections made this way for
@@ -683,7 +683,7 @@ def preprocess_text(textlist, correction_counter=None):
     # to original chunk
     chunk_delimiter = 'chunkdelimiter'
     processed_chunks = [[] for _ in textlist]
-    # clean spacing, normalize case, strip puncutation
+    # clean spacing, normalize case, strip punctuation
     # (temporarily leave punctuation useful for POS tagging)
     full_text = f' {chunk_delimiter} '.join(textlist).lower()
     punc_stripped = re.sub("[^a-zA-Z\s']+", '', full_text.replace('-', ' '))
@@ -783,7 +783,7 @@ def rbf_sum(obs_coords, pred_coords, width, metric='euclidean'):
     width : scalar
         The Width of the Gaussian kernel.
     metric : str or callable, optional
-        The metric used to compute the pairwise distance between
+        The metric used to compute the pairwise distances between
         coordinates (default: `'euclidean'`, Euclidean distance). May be
         any named metric accepted by `scipy.spatial.distance.cdist` or a
         callable that takes two `array_like` arguments.
@@ -806,14 +806,14 @@ def reconstruct_trace(lecture, questions, accuracy):
 
     Parameters
     ----------
-    lecture: numpy.ndarray
+    lecture : numpy.ndarray
         `(n_coordinates, n_features)` matrix of coordinates for which to
         estimate knowledge.
-    questions: numpy.ndarray
+    questions : numpy.ndarray
         `(n_observations, n_features)` matrix of coordinates for the
         quiz questions used to estimate knowledge for each of the
         `n_coordinates` locations.
-    accuracy: array_like
+    accuracy : array_like
         `(n_observations,)` binary array denoting whether each question
         was answered correctly (`True`|`1`) or incorrectly
         (`False`/`0`).
@@ -963,3 +963,5 @@ def z2r(z):
         Correlation value(s).
     """
     return (np.exp(2 * z) - 1) / (np.exp(2 * z) + 1)
+
+

code/khan_helpers/khan_helpers/participant.py

@@ -137,10 +137,16 @@ def get_data(self, lecture=None, quiz=None):
     def get_kmap(self, kmap_key):
         """
         dict.get()-like access to self.knowledge_maps
-        :param trace_key: str
-                The key for the trace to be returned
-        :return: trace: np.ndarray
-                The trace stored under the given `trace_key`
+
+        Parameters
+        ----------
+        kmap_key : str
+            The key for the knowledge map to be returned
+
+        Returns
+        -------
+        kmap : numpy.ndarray
+            The knowledge map stored under the given `kmap_key`
         """
         try:
             return self.knowledge_maps[kmap_key]
@@ -154,10 +160,16 @@ def get_kmap(self, kmap_key):
     def get_trace(self, trace_key):
         """
         dict.get()-like access to self.traces
-        :param trace_key: str
-                The key for the trace to be returned
-        :return: trace: np.ndarray
-                The trace stored under the given `trace_key`
+
+        Parameters
+        ----------
+        trace_key : str
+            The key for the trace to be returned
+
+        Returns
+        -------
+        trace : numpy.ndarray
+            The trace stored under the given `trace_key`
         """
         try:
             return self.traces[trace_key]

code/notebooks/README.md

@@ -1,21 +1,24 @@
 This folder contains notebooks to replicate analyses and figures from the 
-[paper](../../per/main.pdf). Notebooks are generally ordered according to when 
+[paper](../../paper/main.pdf). Notebooks are generally ordered according to when 
 analyses are described or figures appear in the paper.
 
 ### `main/`
 **code to reproduce analyses and figures from the main text**
-- `1_model-course-content.ipynb` – code to fit the topic model, transform lectures & quizzes, and generate Figure 2C
+- `1_model-course-content.ipynb` – code to fit the topic model, transform lectures & quiz questions, and generate Figure 2C
 - `2_topic-variability-fig.ipynb` – code to generate Figure 3 and compute related stats
 - `3_lecture-question-timeseries-fig.ipynb` – code to generate Figure 4
 - `4_reconstructing-knowledge.ipynb` – code to estimate dynamic knowledge traces, generate Figure 5, and compute related stats
 - `5_predictive-analyses.ipynb` – code to generate Figure 6
-- `6_knowledge-maps.ipynb` – code to construct knowledge maps and generate Figure 7
-- `7_parse_demographics.ipynb` – code to compile demographic data described in _Methods_
-- `8_get-transcripts.ipynb` – code to fetch automated transcripts from the YouTube API as described in _Methods_
+- `6_knowledge-smoothness.ipynb` – code to generate Figure 7
+- `7_knowledge-maps.ipynb` – code to construct knowledge maps and generate Figure 8
+- `8_parse_demographics.ipynb` – code to compile demographic data described in _Methods_
+- `9_get-transcripts.ipynb` – code to fetch automated transcripts from the YouTube API as described in _Methods_
 
 ### `supp/`
 **code to reproduce tables and figures from the Supplementary Materials document**
-- `1_tables.ipynb` – code to generate Supplementary Tables 1 & 2
+- `1_tables.ipynb` – code to generate Supplementary Tables 1, 2, and 3, and compute the correlation timeseries peaks shown in Supplementary Figures 3 & 4
 - `2_topic-word-distributions-fig.ipynb` – code to generate Supplementary Figure 1
-- `2_topic-weight-fig.ipynb` – code to generate Supplementary Figure 2 and compute related stats
-- `3_individual-maps.ipynb` – code to generate Supplementary Figures 2–6
+- `3_topic-weight-fig.ipynb` – code to generate Supplementary Figure 2 and compute related stats
+- `4_lecture-question-correlation-peaks.ipynb` – code to generate Supplementary Figures 3 & 4
+- `5_model-comparison-BERT.ipynb` – code to generate Supplementary Figure 5
+- `6_individual-maps.ipynb` – code to generate Supplementary Figures 2–6