Simple translation of Mike's example R script to Python

mcfrank_analysis.py

import pandas as pd
import seaborn as sns
imoprt statsmodels.api as sm

#### 1. read in data
#d <- read.csv("data/all_data.csv") 

d = pd.read_csv("data/all_data.csv")

#### 2. aggregate for each subject and then across subjects
#mss <- aggregate(side ~ subid + agegroup + corr.side + condition, 
#                 data = d, mean)
#ms <- aggregate(side ~ agegroup + corr.side + condition, 
#                data = mss, mean)

ms = d.groupby(["agegroup", "corr_sid", "condition"]).side.mean().reset_index()
mss = ms.groupyby("subid").side.mean().reset_index()

#### 3. plot
#qplot(agegroup, side, colour = corr.side, 
#      facets = .~condition,  
#      group = corr.side, 
#      geom = "line", 
#      data = ms)

sns.factorplot("agegroup", "side", "corr_side", col="condition", data=ms, kind="point")

#### 4. linear mixed-effects model
#lm.all <- glmer(side ~ condition * corr.side * age + 
#                (corr.side | subid), 
#                data = kids, family = "binomial")

# Womp womp, not in Python yet.
# But there is currently a PR in statsmodels with mixed effects regression.
# At this point I would use the IPython rmagic function to run glmer in an R cell
# with very little interruption to the workflow