data_science_in_julia_for_hackers
data_science_in_julia_for_hackers copied to clipboard
Chapter 3 concepts review
In Chapter 3 in the Introduction of conditional probability: It is explained that the rain influences the probability that we stay home. P(H|R) != P(H) Then it is assumed that the rain is not influenced by staying at home or not: P(R|H) = P(R). This might seem logical from a causal point of view but not from a correlation point of view and as my understanding is and you explain, the probabilities don't relate to causality but to correlation. Later in the introduction of Bayes formula, there is even a direct relation between P(H|R) and P(R|H), (likelihood and posterior). It also makes sense that if I always stay at home when it rains, than P(R|H) is also increased. You can infer from me staying home, that it might be raining. So I think this point might need rethinking and rephrasing and maybe a better analogy that is not confusing.
-
Probability Distribution: The explanation is a bit too fuzzy (wrong) maybe. The probability of exactly being 60in tall is 0. The probability of being in the range [59.5;60.5] is about 0.06 I would guess. The approach is good to keep it simple here, but it should be "correct" as well. I guess it is unavoidable to have a distinction between discrete and continuous... but maybe that can be (or is) introduced later on.
-
Normalized Histogram Rainfall in Buenos Aires: The normalization should be a bit more explained. From the explanation, I would think that being in the range 100-120 (one bin) would have the probability of about 0.004. But this is too low by a factor of 20 (size of bins). The probability is normalized so that each 1mm range in the 100-120 range has a probability of 0.004. This is a bit confusing, so I think there might be a better way to normalize or explain it more in detail what the normalization does and how one would sum it so that 1 comes out. When I see probabilities I always want to sum them to 1, here it is a bit difficult to see that 1 comes out.
-
First histogram Julia code to convert from "februero" to "february". Wouldn't it be nicer to store the translation in a dictionary and translate it