bledem / RunAnova

Package to run one way and two way anova with practical feature to understand the distrubution of the sample data.

RunAnova

This library run ANOVA one way and two way experiments. RunAnova will detail the computation steps (Mean Square, Degree of Freedom), to help you understand the results.

Install the package (optinal)

If you want to install the package in your environment:

conda create --name anova python=3.7
pip install dist/RunAnova-1.0a0-py3-none-any.whl

Run example notebooks

One way using RunAnova

• example_oneway_breakfast_test.ipynb

One way using classic packages

• example_without_library.ipynb

Two ways using RunAnova

• two_ways_test.ipynb

Results

Example of results for one way configuration:

For the following example dataframe:

No breakfast	Light breakfast	Full breaskfast
8	14	10
7	16	12
9	12	16
13	17	15
10	11	12

Init and call a OneWayAnova object:

from RunAnova import OneWayAnova
resp_var = 'attention_span'
treatments = "meal_quantity"
anov = OneWayAnova(df=exp_df, resp_var=resp_var , treatments=treatments)
anov(simple=True)
anov(simple=False)

The output of ANOVA is either simple:

F stat: 4.9, p-value: 0.03

or detailed:

##### Decomposition of variance #####
Grand Mean 12.13
SST: 58.53, dof k-1:  2, MST: 29.27
SSE: 58.53, dof n-k:  12, MSE: 5.93
TSS=SST+SSE total variation: 129.73, DOF: 14
##### Testing #####
Test statistic: F=MST/MSE=4.93
F_alpha: 3.89 for a level of confidence of 0.05
We reject Ho!
We failed to reject Ho: one mean is different from the others!
P-value: 0.0273
eta squared: 0.45 omega_squared: 58.45

Example of results for two way configuration:

Init and call a TwoWayAnova class object:

from RunAnova import TwoWayAnova
factors_levels = {'wheel':[], 'gas':[]}
df = pd.DataFrame({'wheel': np.tile(['two', 'four'], 4),
                   'gas': np.tile(['regular', 'octane', 'octane', 'regular'],2),
                   'consumption': [26.7, 26.1, 32.3, 28.6,
                          25.2, 24.2, 32.8, 29.3]})      
for col in df.columns:
    if col in factors_levels:
        factors_levels[col] = list(df[col].unique())
anova_proc = TwoWayAnova(df, resp_var='consumption', factors_levels=factors_levels)
anova_proc()

##### Decomposition of variance #####
Grand Mean 12.13
SST: 58.53, dof k-1:  2, MST: 29.27
SSE: 58.53, dof n-k:  12, MSE: 5.93
TSS=SST+SSE total variation: 129.73, DOF: 14
##### Testing #####
Test statistic: F=MST/MSE=4.93
F_alpha: 3.89 for a level of confidence of 0.05
We reject Ho!
We failed to reject Ho: one mean is different from the others!
P-value: 0.0273
eta squared: 0.45 omega_squared: 58.45