@@ -74,3 +74,19 @@ In this exercise, I performed a simple statistical analysis. Using the provided
...
@@ -74,3 +74,19 @@ In this exercise, I performed a simple statistical analysis. Using the provided
Based on the dataset provided in the previous excercise (2.2), I reproduced a **sequence plot** and **a histogram** to visualize the data.
Based on the dataset provided in the previous excercise (2.2), I reproduced a **sequence plot** and **a histogram** to visualize the data.
## Exercise 02 (5th part): Critical examination of a data analysis (Challenger O-ring case)
In this task, I worked with a historical dataset related to the **risk of failure of the O-ring seals** in the space shuttle *Challenger*. This dataset gained significance after the tragic explosion of the shuttle 73 seconds after launch, resulting in the loss of seven astronauts.
Prior to the launch, a teleconference was held to assess the risk of taking off on an exceptionally cold morning. However, despite discussions, the flight was not delayed, and the analysis presented at the time significantly underestimated the probability of failure. My objective was to critically review the analysis and identify the major reasoning errors that led to the flawed conclusion.
**Key steps taken:**
- I reviewed the historical context and the principles of logistic regression, using the resource provided in GitLab.
- I analyzed the dataset and evaluated the model that was used at the time.
- I identified three majore errors that caused the underestimation of failure probability:
1.**No data was available for low temperatures**, yet conclusions were still drawn.
2.**Key data points were excluded** from the analysis.
3.**the uncertainity of the logistic regression estimates was not taken into accounts**, leading to overconfidence in the predictions.