"from all angles in order to to explain what's wrong."
"from all angles in order to to explain what's wrong."
]
]
},
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Problem: the above analysis omitted a possible confounding variable (Pressure)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We see in the examples below that 1) pressure and temperature are related, and 2) very low or very high pressure may increase frequency of malfunction."
"We see that the sign of the temperature effect, although still not significant (likely due to the small number of data), is reversed compared to the non-adjusted analysis. Now, low temperature are associated with increased frequencies of malfunctions."
"### Conclusion: by adjusting for a possible confounder, the malfunction frequency is predicted to be much higher than in the non-adjusted model (between 0.5 and 0.8 depending on the pressure, compared to 0.2 in the naive analysis)."
