Essential Statistics for DS - 1
Topics: 1. Log Scale and Usage How many times a “number” should be multiplied by itself to reach a certain “target”. Change growth/existence of a quantity/number with respect to base 10. how many times 10 should be multiplied with 10. Thus a relative value to 10. 2. Mean Absolute Deviation Mean Absolute Deviation : First compute absolute deviation of all data points from mean value. Then take mean of these deviation values. Its MAD. Mean of all absolute deviations. If MAD > means spread is wide i.e. most data points are located at good distance from mean value. If MAD < means spread is narrow i.e. most data points are located near to mean value. 3. Need to Standard Deviation Its observed in many cases that MAD > or < can not predict actual spread of data points from mean. In most cases where outliers exist, it has been observed that MAD does not provide good details of spread. To overcome this process, there must be some process which penalize such ou...