What a Drunk Man Can Teach Us About Time Series Forecasting
\n \t
\n
<\/BR>
\n
\n \t
\n
<\/BR><\/p>\n
Autocorrelation & The Random Walk explained<\/strong> with a drunk man \ud83c\udf7a<\/p>\n Let me illustrate this statistical concept with an example we can all visualize.<\/p>\n Imagine a drunk man wandering a city. His steps are completely random and unpredictable.<\/p>\n Here’s the intuition<\/strong>:<\/p>\n – His current position is completely tied to his previous position<\/p>\n – We know where he is RIGHT NOW, but have no idea where he’ll be in the next minute<\/p>\n The statistical insight<\/strong>:<\/p>\n In a random walk, the current position<\/em> is highly correlated with the previous position<\/em>, but the changes in position<\/em> (the steps) are completely random & uncorrelated.<\/p>\n This is why random walks are so tricky to forecast!<\/p>\n Part 2: Time Series Forecasting: Build a Baseline & Understand the Random Walk<\/a> <\/p>\n Would love to hear your thoughts, feedback<\/strong> about this topic<\/p>\n<\/p><\/div>\n submitted by \/u\/The_Simpsons_22 <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span>\n<\/div>\n