def decision_tree_tests():
xs1 = np.array([1,1,1,1,1,1])
xs2 = np.array([1,1,0,0,1,0])
xs3 = np.array([1,0,1,0,1,0])
xs4 = np.array([1,0,0,0,1,0])
ys = np.array([1,1,0,0,1,0])
ent_base = dt.entropy_calc(ys,[],[])
ent_x1 = dt.entropy_calc(ys,xs1,[0,1])
ent_x2 = dt.entropy_calc(ys,xs2,[0,1])
ent_x3 = dt.entropy_calc(ys,xs3,[0,1])
ent_x4 = dt.entropy_calc(ys,xs4,[0,1])
ent_x3_2 = dt.entropy_calc(ys,xs3,[0,1,2])
if ent_base != 1.0:
print 'ERROR: Base case for entropy calculation incorrect.'
if ent_x1 != 1.0:
print 'Error: Random data should give entropy = 1!'
print 'Calculated entropy is ' + str(ent_x1)
if ent_x2 != 0.0:
print 'Error: Perfect data should give entropy = 0!'
print 'Calculated entropy is ' + str(ent_x2)
if ent_x3 < ent_x2:
print 'Error: Imperfect data should do worse than perfect data!'
print 'Malformed expression: ' + str(ent_x3) + '<' + str(ent_x2)
if ent_x3 > ent_x1:
print 'Error: Imperfect data should do better than random data!'
print 'Malformed expression: ' + str(ent_x3) + '>' + str(ent_x1)
if ent_x3 < ent_x4:
print 'Error: Malformed expression: ' + str(ent_x3) + '<' + str(ent_x4)
if ent_x3_2 != ent_x3:
print 'Error: Empty class should not matter'
tag_master = tg.string_tags(xy_train['KC(Default)'])
[tag_array,opp_array] = tg.tags_to_array(xy_train['KC(Default)'],xy_train['Opportunity(Default)'],tag_master)
#Look up location of index in array
#tag_master.index(SOME STRING)
int_s = ['Correct First Attempt','Incorrects','Hints','Corrects']
for i in range(len(int_s)):
xy_train[int_s[i]] = map(int,xy_train[int_s[i]])
y_pred = xy_train['Correct First Attempt']
#Check entropy of the data
ent = dt.entropy_calc(y_pred,[0],[])
def step_normalize(stud_IDs,stud_dict,step_start_time,first_trans_time,corr_trans_time,step_end_time):
# Normalizes the step start time by student's first transation time
aa = np.copy(step_start_time)
bb = np.copy(first_trans_time)
dd = np.copy(step_end_time)
cc = np.copy(corr_trans_time)
for stud in stud_dict:
print('Processing student ' + str(stud))
rel_steps = [step_start_time[i] for i in np.where(stud_IDs == stud)][0]
rel_ind = np.where(stud_IDs == stud)[0]
# In case this array isn't sorted...
rel_steps_ind_sort = np.argsort(rel_steps)
fnz = [i for i in rel_steps if i > 0]