Interactions and ANOVA¶
Note: This script is based heavily on Jonathan Taylor's class notes http://www.stanford.edu/class/stats191/interactions.html
Download and format data:
In [1]:
%matplotlib inline
from __future__ import print_function
from statsmodels.compat import urlopen
import numpy as np
np.set_printoptions(precision=4, suppress=True)
import statsmodels.api as sm
import pandas as pd
pd.set_option("display.width", 100)
import matplotlib.pyplot as plt
from statsmodels.formula.api import ols
from statsmodels.graphics.api import interaction_plot, abline_plot
from statsmodels.stats.anova import anova_lm
try:
salary_table = pd.read_csv('salary.table')
except: # recent pandas can read URL without urlopen
url = 'http://stats191.stanford.edu/data/salary.table'
fh = urlopen(url)
salary_table = pd.read_table(fh)
salary_table.to_csv('salary.table')
E = salary_table.E
M = salary_table.M
X = salary_table.X
S = salary_table.S
Take a look at the data:
In [2]:
plt.figure(figsize=(6,6))
symbols = ['D', '^']
colors = ['r', 'g', 'blue']
factor_groups = salary_table.groupby(['E','M'])
for values, group in factor_groups:
i,j = values
plt.scatter(group['X'], group['S'], marker=symbols[j], color=colors[i-1],
s=144)
plt.xlabel('Experience');
plt.ylabel('Salary');
Fit a linear model:
In [3]:
formula = 'S ~ C(E) + C(M) + X'
lm = ols(formula, salary_table).fit()
print(lm.summary())
Have a look at the created design matrix:
In [4]:
lm.model.exog[:5]
Out[4]:
Or since we initially passed in a DataFrame, we have a DataFrame available in
In [5]:
lm.model.data.orig_exog[:5]
Out[5]:
We keep a reference to the original untouched data in
In [6]:
lm.model.data.frame[:5]
Out[6]:
Influence statistics
In [7]:
infl = lm.get_influence()
print(infl.summary_table())
or get a dataframe
In [8]:
df_infl = infl.summary_frame()
In [9]:
df_infl[:5]
Out[9]:
Now plot the reiduals within the groups separately:
In [10]:
resid = lm.resid
plt.figure(figsize=(6,6));
for values, group in factor_groups:
i,j = values
group_num = i*2 + j - 1 # for plotting purposes
x = [group_num] * len(group)
plt.scatter(x, resid[group.index], marker=symbols[j], color=colors[i-1],
s=144, edgecolors='black')
plt.xlabel('Group');
plt.ylabel('Residuals');
Now we will test some interactions using anova or f_test
In [11]:
interX_lm = ols("S ~ C(E) * X + C(M)", salary_table).fit()
print(interX_lm.summary())
Do an ANOVA check
In [12]:
from statsmodels.stats.api import anova_lm
table1 = anova_lm(lm, interX_lm)
print(table1)
interM_lm = ols("S ~ X + C(E)*C(M)", data=salary_table).fit()
print(interM_lm.summary())
table2 = anova_lm(lm, interM_lm)
print(table2)
The design matrix as a DataFrame
In [13]:
interM_lm.model.data.orig_exog[:5]
Out[13]:
The design matrix as an ndarray
In [14]:
interM_lm.model.exog
interM_lm.model.exog_names
Out[14]:
In [15]:
infl = interM_lm.get_influence()
resid = infl.resid_studentized_internal
plt.figure(figsize=(6,6))
for values, group in factor_groups:
i,j = values
idx = group.index
plt.scatter(X[idx], resid[idx], marker=symbols[j], color=colors[i-1],
s=144, edgecolors='black')
plt.xlabel('X');
plt.ylabel('standardized resids');
Looks like one observation is an outlier.
In [16]:
drop_idx = abs(resid).argmax()
print(drop_idx) # zero-based index
idx = salary_table.index.drop(drop_idx)
lm32 = ols('S ~ C(E) + X + C(M)', data=salary_table, subset=idx).fit()
print(lm32.summary())
print('\n')
interX_lm32 = ols('S ~ C(E) * X + C(M)', data=salary_table, subset=idx).fit()
print(interX_lm32.summary())
print('\n')
table3 = anova_lm(lm32, interX_lm32)
print(table3)
print('\n')
interM_lm32 = ols('S ~ X + C(E) * C(M)', data=salary_table, subset=idx).fit()
table4 = anova_lm(lm32, interM_lm32)
print(table4)
print('\n')
Replot the residuals
In [17]:
try:
resid = interM_lm32.get_influence().summary_frame()['standard_resid']
except:
resid = interM_lm32.get_influence().summary_frame()['standard_resid']
plt.figure(figsize=(6,6))
for values, group in factor_groups:
i,j = values
idx = group.index
plt.scatter(X[idx], resid[idx], marker=symbols[j], color=colors[i-1],
s=144, edgecolors='black')
plt.xlabel('X[~[32]]');
plt.ylabel('standardized resids');
Plot the fitted values
In [18]:
lm_final = ols('S ~ X + C(E)*C(M)', data = salary_table.drop([drop_idx])).fit()
mf = lm_final.model.data.orig_exog
lstyle = ['-','--']
plt.figure(figsize=(6,6))
for values, group in factor_groups:
i,j = values
idx = group.index
plt.scatter(X[idx], S[idx], marker=symbols[j], color=colors[i-1],
s=144, edgecolors='black')
# drop NA because there is no idx 32 in the final model
plt.plot(mf.X[idx].dropna(), lm_final.fittedvalues[idx].dropna(),
ls=lstyle[j], color=colors[i-1])
plt.xlabel('Experience');
plt.ylabel('Salary');
From our first look at the data, the difference between Master's and PhD in the management group is different than in the non-management group. This is an interaction between the two qualitative variables management,M and education,E. We can visualize this by first removing the effect of experience, then plotting the means within each of the 6 groups using interaction.plot.
In [19]:
U = S - X * interX_lm32.params['X']
plt.figure(figsize=(6,6))
interaction_plot(E, M, U, colors=['red','blue'], markers=['^','D'],
markersize=10, ax=plt.gca())
Out[19]:
Minority Employment Data¶
In [20]:
try:
jobtest_table = pd.read_table('jobtest.table')
except: # don't have data already
url = 'http://stats191.stanford.edu/data/jobtest.table'
jobtest_table = pd.read_table(url)
factor_group = jobtest_table.groupby(['MINORITY'])
fig, ax = plt.subplots(figsize=(6,6))
colors = ['purple', 'green']
markers = ['o', 'v']
for factor, group in factor_group:
ax.scatter(group['TEST'], group['JPERF'], color=colors[factor],
marker=markers[factor], s=12**2)
ax.set_xlabel('TEST');
ax.set_ylabel('JPERF');
In [21]:
min_lm = ols('JPERF ~ TEST', data=jobtest_table).fit()
print(min_lm.summary())
In [22]:
fig, ax = plt.subplots(figsize=(6,6));
for factor, group in factor_group:
ax.scatter(group['TEST'], group['JPERF'], color=colors[factor],
marker=markers[factor], s=12**2)
ax.set_xlabel('TEST')
ax.set_ylabel('JPERF')
fig = abline_plot(model_results = min_lm, ax=ax)
In [23]:
min_lm2 = ols('JPERF ~ TEST + TEST:MINORITY',
data=jobtest_table).fit()
print(min_lm2.summary())
In [24]:
fig, ax = plt.subplots(figsize=(6,6));
for factor, group in factor_group:
ax.scatter(group['TEST'], group['JPERF'], color=colors[factor],
marker=markers[factor], s=12**2)
fig = abline_plot(intercept = min_lm2.params['Intercept'],
slope = min_lm2.params['TEST'], ax=ax, color='purple');
fig = abline_plot(intercept = min_lm2.params['Intercept'],
slope = min_lm2.params['TEST'] + min_lm2.params['TEST:MINORITY'],
ax=ax, color='green');
In [25]:
min_lm3 = ols('JPERF ~ TEST + MINORITY', data = jobtest_table).fit()
print(min_lm3.summary())
In [26]:
fig, ax = plt.subplots(figsize=(6,6));
for factor, group in factor_group:
ax.scatter(group['TEST'], group['JPERF'], color=colors[factor],
marker=markers[factor], s=12**2)
fig = abline_plot(intercept = min_lm3.params['Intercept'],
slope = min_lm3.params['TEST'], ax=ax, color='purple');
fig = abline_plot(intercept = min_lm3.params['Intercept'] + min_lm3.params['MINORITY'],
slope = min_lm3.params['TEST'], ax=ax, color='green');
In [27]:
min_lm4 = ols('JPERF ~ TEST * MINORITY', data = jobtest_table).fit()
print(min_lm4.summary())
In [28]:
fig, ax = plt.subplots(figsize=(8,6));
for factor, group in factor_group:
ax.scatter(group['TEST'], group['JPERF'], color=colors[factor],
marker=markers[factor], s=12**2)
fig = abline_plot(intercept = min_lm4.params['Intercept'],
slope = min_lm4.params['TEST'], ax=ax, color='purple');
fig = abline_plot(intercept = min_lm4.params['Intercept'] + min_lm4.params['MINORITY'],
slope = min_lm4.params['TEST'] + min_lm4.params['TEST:MINORITY'],
ax=ax, color='green');
In [29]:
# is there any effect of MINORITY on slope or intercept?
table5 = anova_lm(min_lm, min_lm4)
print(table5)
In [30]:
# is there any effect of MINORITY on intercept
table6 = anova_lm(min_lm, min_lm3)
print(table6)
In [31]:
# is there any effect of MINORITY on slope
table7 = anova_lm(min_lm, min_lm2)
print(table7)
In [32]:
# is it just the slope or both?
table8 = anova_lm(min_lm2, min_lm4)
print(table8)
One-way ANOVA¶
In [33]:
try:
rehab_table = pd.read_csv('rehab.table')
except:
url = 'http://stats191.stanford.edu/data/rehab.csv'
rehab_table = pd.read_table(url, delimiter=",")
rehab_table.to_csv('rehab.table')
fig, ax = plt.subplots(figsize=(8,6))
fig = rehab_table.boxplot('Time', 'Fitness', ax=ax, grid=False)
In [34]:
rehab_lm = ols('Time ~ C(Fitness)', data=rehab_table).fit()
table9 = anova_lm(rehab_lm)
print(table9)
print(rehab_lm.model.data.orig_exog)
In [35]:
print(rehab_lm.summary())
Two-way ANOVA¶
In [36]:
try:
kidney_table = pd.read_table('./kidney.table')
except:
url = 'http://stats191.stanford.edu/data/kidney.table'
kidney_table = pd.read_csv(url, delim_whitespace=True)
Explore the dataset
In [37]:
kidney_table.head(10)
Out[37]:
Balanced panel
In [38]:
kt = kidney_table
plt.figure(figsize=(8,6))
fig = interaction_plot(kt['Weight'], kt['Duration'], np.log(kt['Days']+1),
colors=['red', 'blue'], markers=['D','^'], ms=10, ax=plt.gca())
You have things available in the calling namespace available in the formula evaluation namespace
In [39]:
kidney_lm = ols('np.log(Days+1) ~ C(Duration) * C(Weight)', data=kt).fit()
table10 = anova_lm(kidney_lm)
print(anova_lm(ols('np.log(Days+1) ~ C(Duration) + C(Weight)',
data=kt).fit(), kidney_lm))
print(anova_lm(ols('np.log(Days+1) ~ C(Duration)', data=kt).fit(),
ols('np.log(Days+1) ~ C(Duration) + C(Weight, Sum)',
data=kt).fit()))
print(anova_lm(ols('np.log(Days+1) ~ C(Weight)', data=kt).fit(),
ols('np.log(Days+1) ~ C(Duration) + C(Weight, Sum)',
data=kt).fit()))
Sum of squares¶
Illustrates the use of different types of sums of squares (I,II,II) and how the Sum contrast can be used to produce the same output between the 3.
Types I and II are equivalent under a balanced design.
Don't use Type III with non-orthogonal contrast - ie., Treatment
In [40]:
sum_lm = ols('np.log(Days+1) ~ C(Duration, Sum) * C(Weight, Sum)',
data=kt).fit()
print(anova_lm(sum_lm))
print(anova_lm(sum_lm, typ=2))
print(anova_lm(sum_lm, typ=3))
In [41]:
nosum_lm = ols('np.log(Days+1) ~ C(Duration, Treatment) * C(Weight, Treatment)',
data=kt).fit()
print(anova_lm(nosum_lm))
print(anova_lm(nosum_lm, typ=2))
print(anova_lm(nosum_lm, typ=3))