def gc_scatter_correlation_matrix(raw_dataframe, diagonal='kde',
annotate_correlation=False):
#
# Filter non-numeric columns
# Empty columns will also be dropped.
numeric_colnames = []
for colname in raw_dataframe.columns:
col = dat[colname].dropna()
if (len(col) > 0) and (col.dtype == 'float64'):
numeric_colnames.append(colname)
dframe = raw_dataframe.reindex(columns=numeric_colnames)
colnames = dframe.columns
num_cols = len(colnames)
#
# Find grade ranges (0 to max mark)
grade_ranges = []
for colname in dframe.columns:
max_grade = re.match("^.*\[Total Pts:( up to)? (\d+).*$", colname).group(2)
max_grade = float(max_grade)
grade_ranges.append((0,max_grade))
#
# Rename to filter out GradeCentre chaff
def transform(s):
return re.match("^(.*?)( \[.*$)", s).group(1)
dframe = dframe.rename(columns=transform)
#
# Rename columns with same names
cnts = collections.Counter(dframe.columns)
dupes = {k:v for k,v in cnts.iteritems() if v>=2}
for name, _ in dupes.iteritems():
dupe_indxs = dframe.columns.get_loc(name).nonzero()[0] # indexes of duplicates
i = 0
for indx in dupe_indxs:
i +=1
prev_cols = list(dframe.columns)
prev_cols[indx] = "%s (%d)" % (name, i)
dframe.columns = prev_cols
#
# Create grade range dictionary
ranges_dict = {}
for indx, colname in enumerate(dframe.columns):
ranges_dict = grade_ranges[indx]
fig = scatter_correlation_matrix(dframe, diagonal=diagonal,
annotate_correlation=annotate_correlation,
ranges=ranges_dict)
return fig
def scm_for_worksheet(df_raw,
diagonal='kde',
annotate_correlation=True,
colour_correlation=True):
"""
df_raw: dataframe representing worksheet for a particular module.
"""
cols = df_raw.columns
ncols = len(cols)
# Obtain subset of cols
indxs = [2] + range(4, ncols)
# 2 = module marks (weighted total)
# range(4,ncols) = individual courseworks
df = df_raw.iloc[:, indxs] #iloc: = 0, n-1
# Drop first row ("Mark", ...)
#df = df.drop([0])
df = df.iloc[1:, :]
df = df.reset_index()
# Drop mystical 'index' col
df = df.drop('index', 1)
# Change dtypes
for cname in df.columns:
df[cname] = df[cname].astype(float)
# Set grade ranges for each coursework
# (Assume all webmark grades are always 0 to 100)
ranges_dict = {}
for cname in df.columns:
ranges_dict[cname] = (0.0, 100.0)
# Plot
fig = scatter_correlation_matrix(df,
diagonal=diagonal,
annotate_correlation=annotate_correlation,
colour_correlation=colour_correlation,
ranges=ranges_dict)
return fig
def scm_for_worksheet(df_raw, diagonal='kde', annotate_correlation=True,
colour_correlation=True):
"""
df_raw: dataframe representing worksheet for a particular module.
"""
cols = df_raw.columns
ncols = len(cols)
# Obtain subset of cols
indxs = [2] + range(4, ncols)
# 2 = module marks (weighted total)
# range(4,ncols) = individual courseworks
df = df_raw.iloc[:,indxs] #iloc: = 0, n-1
# Drop first row ("Mark", ...)
#df = df.drop([0])
df = df.iloc[1:,:]
df = df.reset_index()
# Drop mystical 'index' col
df = df.drop('index', 1)
# Change dtypes
for cname in df.columns:
df[cname] = df[cname].astype(float)
# Set grade ranges for each coursework
# (Assume all webmark grades are always 0 to 100)
ranges_dict = {}
for cname in df.columns:
ranges_dict[cname] = (0.0, 100.0)
# Plot
fig = scatter_correlation_matrix(df, diagonal=diagonal,
annotate_correlation=annotate_correlation,
colour_correlation=colour_correlation,
ranges=ranges_dict)
return fig