# -*- coding: utf-8 -*- # -*- mode: org -*- #+TITLE: Challenger - Python - Emacs - Windows 7 64 bits * Risk Analysis of the Space Shuttle: Pre-Challenger Prediction of Failure In this document we reperform some of the analysis provided in /Risk Analysis of the Space Shuttle: Pre-Challenger Prediction of Failure/ by /Siddhartha R. Dalal, Edward B. Fowlkes, Bruce Hoadley/ published in /Journal of the American Statistical Association/, Vol. 84, No. 408 (Dec., 1989), pp. 945-957 and available at http://www.jstor.org/stable/2290069. On the fourth page of this article, they indicate that the maximum likelihood estimates of the logistic regression using only temperature are: *$\hat{\alpha}$ = 5.085* and *$\hat{\beta}$ = -0.1156* and their asymptotic standard errors are *$s_{\hat{\alpha}}$ = 3.052* and *$s_{\hat{\beta}}$ = 0.047*. The Goodness of fit indicated for this model was *$G^2$ = 18.086* with *21* degrees of freedom. Our goal is to reproduce the computation behind these values and the Figure 4 of this article, possibly in a nicer looking way. ** Technical information on the computer on which the analysis is run We will be using the Python 3 language using the pandas, statsmodels, and numpy library. #+begin_src python :results output :session :exports both def print_imported_modules(): import sys for name, val in sorted(sys.modules.items()): if(hasattr(val, '__version__')): print(val.__name__, val.__version__) # else: # print(val.__name__, "(unknown version)") def print_sys_info(): import sys import platform print(sys.version) print(platform.uname()) import numpy as np import pandas as pd import matplotlib.pyplot as plt import statsmodels.api as sm import seaborn as sns print_sys_info() print_imported_modules() #+end_src ** Loading and inspecting data Let's start by reading data. #+begin_src python :results output :session :exports both data = pd.read_csv("https://app-learninglab.inria.fr/moocrr/gitlab/moocrr-session3/moocrr-reproducibility-study/tree/master/data/shuttle.csv") print(data) #+end_src We know from our previous experience on this data set that filtering data is a really bad idea. We will therefore process it as such. #+begin_src python :results file :session :var matplot_lib_filename=(org-babel-temp-file "figure" ".png") :exports both %matplotlib inline pd.set_option('mode.chained_assignment',None) # this removes a useless warning from pandas data["Frequency"]=data.Malfunction/data.Count data.plot(x="Temperature",y="Frequency",kind="scatter",ylim=[0,1]) plt.grid(True) plt.tight_layout() plt.savefig(matplot_lib_filename) matplot_lib_filename #+end_src ** Logistic regression Let's assume O-rings independently fail with the same probability which solely depends on temperature. A logistic regression should allow us to estimate the influence of temperature. #+begin_src python :results output :session :exports both import statsmodels.api as sm data["Success"]=data.Count-data.Malfunction data["Intercept"]=1 logmodel=sm.GLM(data['Frequency'], data[['Intercept','Temperature']], family=sm.families.Binomial(sm.families.links.logit)).fit() print(logmodel.summary()) #+end_src The maximum likelyhood estimator of the intercept and of Temperature are thus *$\hat{\alpha}$ = 5.0850* and *$\hat{\beta}$ = -0.1156*. This *corresponds* to the values from the article of Dalal /et al./ The standard errors are /$s_{\hat{\alpha}}$ = 7.477/ and /$s_{\hat{\beta}}$ = 0.115/, which is *different* from the *3.052* and *0.04702* reported by Dallal /et al./ The deviance is /3.01444/ with *21* degrees of freedom. I cannot find any value similar to the Goodness of fit (*$G^2$ = 18.086*) reported by Dalal /et al./ There seems to be something wrong. Oh I know, I haven't indicated that my observations are actually the result of 6 observations for each rocket launch. Let's indicate these weights (since the weights are always the same throughout all experiments, it does not change the estimates of the fit but it does influence the variance estimates). #+begin_src python :results output :session :exports both logmodel=sm.GLM(data['Frequency'], data[['Intercept','Temperature']], family=sm.families.Binomial(sm.families.links.logit), var_weights=data['Count']).fit() print(logmodel.summary()) #+end_src Good, now I have recovered the asymptotic standard errors *$s_{\hat{\alpha}}$ = 3.052* and *$s_{\hat{\beta}}$ = 0.047*. The Goodness of fit (Deviance) indicated for this model is *$G^2$ = 18.086* with *21* degrees of freedom (Df Residuals). *I have therefore managed to fully replicate the results of the Dalal /et al./ article*. ** Predicting failure probability The temperature when launching the shuttle was 31°F. Let's try to estimate the failure probability for such temperature using our model: #+begin_src python :results output :session :exports both data_pred = pd.DataFrame({'Temperature': np.linspace(start=30, stop=90, num=121), 'Intercept': 1}) data_pred['Frequency'] = logmodel.predict(data_pred) print(data_pred.head()) #+end_src #+begin_src python :results file :session :var matplot_lib_filename=(org-babel-temp-file "figure" ".png") :exports both %matplotlib inline data_pred.plot(x="Temperature",y="Frequency",kind="line",ylim=[0,1]) plt.scatter(x=data["Temperature"],y=data["Frequency"]) plt.grid(True) plt.savefig(matplot_lib_filename) matplot_lib_filename #+end_src La fonction =logmodel.predict(data_pred)= ne fonctionne pas avec les dernières versions de pandas (Frequency = 1 pour toutes les températures). On peut alors utiliser le code suivant pour calculer les prédictions et tracer la courbe : #+begin_src python :results output :session :exports both # Inspiring from http://blog.yhat.com/posts/logistic-regression-and-python.html def logit_inv(x): return(np.exp(x)/(np.exp(x)+1)) data_pred['Prob']=logit_inv(data_pred['Temperature'] * logmodel.params['Temperature'] + logmodel.params['Intercept']) print(data_pred.head()) #+end_src #+begin_src python :results file :session :var matplot_lib_filename=(org-babel-temp-file "figure" ".png") :exports both %matplotlib inline data_pred.plot(x="Temperature",y="Prob",kind="line",ylim=[0,1]) plt.scatter(x=data["Temperature"],y=data["Frequency"]) plt.grid(True) plt.savefig(matplot_lib_filename) matplot_lib_filename #+end_src This figure is very similar to the Figure 4 of Dalal /et al./ *I have managed to replicate the Figure 4 of the Dalal /et al./ article.* ** Computing and plotting uncertainty Following the documentation of [Seaborn](https://seaborn.pydata.org/generated/seaborn.regplot.html), I use regplot. #+begin_src python :results file :session :var matplot_lib_filename=(org-babel-temp-file "figure" ".png") :exports both sns.set(color_codes=True) plt.xlim(30,90) plt.ylim(0,1) sns.regplot(x='Temperature', y='Frequency', data=data, logistic=True) plt.show() plt.savefig(matplot_lib_filename) matplot_lib_filename #+end_src **I think I have managed to correctly compute and plot the uncertainty of my prediction.** Although the shaded area seems very similar to [the one obtained by with R](https://app-learninglab.inria.fr/moocrr/gitlab/moocrr-session3/moocrr-reproducibility-study/tree/master/challenger.pdf), I can spot a few differences (e.g., the blue point for temperature 63 is outside)... Could this be a numerical error ? Or a difference in the statistical method ? It is not clear which one is "right".