Data-based decision-making - the process of data-based decision-making is a quantitative process Data-based decision-making is to assist us in decision-making through quantitative data, thereby improving the scientifi city and accuracy of decision-making. 1. Understand quantification The earliest scientists in history did not admit that experiments could have errors. They believed that all measurements must be accurate, and attributed any errors to errors. It was not until later that people gradually realized that errors will always exist and cannot be eliminated. The same is true of quantification. Quantification is to reduce uncertainty and estimate risks, so as to assist decision-making.
Therefore, the process of quantification does not need to be infinitely accurate, and it does not need to completely eliminate uncertainty, as long as it can support our decision-making. 2. Confidence intervals - a way of quantification Because quantification is not necessarily an exact number, and in reality, we often encounter imperfect data, and the amount mobile number list of data is too large to be processed in a short period of time, so we introduce a statistical concept - confidence interval, to assist We decide. Confidence intervals are the ranges that represent a correct answer with a specific probability. In general, we require the confidence interval to be narrow enough and the confidence to be above 80%.
If the confidence is too low, it means that the data interval is very likely to be wrong, and if the interval is too large, it means that the interval lacks reference significance. For example: 100% confidence in the score of this test is [0, 100], this interval is equal to nothing, and there is no reference significance; the score of this test has 5% confidence in [95, 100], which means that this time The test scores are 95% confident in [0, 95], so the interval [95, 100] is probably wrong. The score of this test has 80% confidence in [85,100], which means that this range is likely to be correct and can reflect the real situation, and even we can think that the average score of the class is around 92.5. exclusive! How to play data analysis? Confidence Interval Example 2.
Data disassembly 1. Identify goals - goals must be quantifiable Each project has a certain goal, so before we do it, we must understand what our goal is. Sometimes, the business or product will directly tell us what the goal is, such as improving the retention rate and improving the conversion rate. At this time, the goal It is very clear, we can directly dismantle the target. Of course, sometimes the target will be vague, such as improving the user experience. At this time, we need to make the target quantifiable by clarifying the chain. 2. Clarification Chain - Make the goal quantifiable A chain of clarification is a series of short links that imagine something as intangible to tangible. For example, sometimes our goal is to improve the user experience. This goal is not in line with the measurable item in the "SMART principle".