2c:null 2d:["$","title",null,{"children":"How Many Samples Do I Need? | Troy Magennis | Observable"}] 32:T495,md`And so on. This of course assumes a few things that aren't always tue and probably matter. 1. That all areas between the samples are equally spaced, they won't be. 2. That the probability of each segment is equal, they won't be. 3. That there are no duplicates, there will be 4. That the sampels are gathered at random, they won't be (the next week isn't random, its just the next week!) But, the basic condition that will always be true is that you get the most information from the first few samples then the new information given by each sample tapers off. ## The question you really want answered is: How many samples do I need, just tell me. 7 to 15 And now in consultant tense: It depends. If the data is changing and recency matters (team size changes, new frameworks), go for lower sample counts around 7. If the process is pretty stable (similar team sizes, similar work types) then go up to 15 samples. Always look for reasons why samples should be deemed invalid for future forecasting. Overtime and late crunching delivery pace is a killer. It makes the team look to be delivering faster than they can, and embeds that overtime into future projects! `33:[] 30:{"id":"3ac9ca98c6f349ee","slug":"how-many-samples-do-i-need","title":"How Many Samples Do I Need?","thumbnail":"e97581df67530394769010b124d0dcaaf6cd12af69808aef5336efb66607b4c4","access_level":"public","create_time":"$D2021-10-12T18:00:17.902Z","update_time":"$D2021-10-25T22:42:36.078Z","view_time":null,"first_public_time":"$D2021-10-19T21:22:25.357Z","has_unpublished_edits":null,"fork":false,"stars":5,"starred":false,"owner":{"type":"team","id":"5bfcd1f236454eeb","login":"troymagennis","name":"Troy Magennis","avatar_url":"https://avatars-next.observableusercontent.com/bd84c523cb40d1d089b9d8d0a4ac5e025e3593ce8554df5d578d6da6190496c5","tier":"starter_2024"},"version":152,"publish_version":152,"role":null,"authored":false,"authors":[{"type":"individual","id":"f950cac864488237","login":"troymagennis","name":"Troy Magennis","avatar_url":"https://avatars-next.observableusercontent.com/bd84c523cb40d1d089b9d8d0a4ac5e025e3593ce8554df5d578d6da6190496c5","tier":null}],"members":[{"type":"individual","id":"f950cac864488237","login":"troymagennis","name":"Troy Magennis","avatar_url":"https://avatars-next.observableusercontent.com/bd84c523cb40d1d089b9d8d0a4ac5e025e3593ce8554df5d578d6da6190496c5","tier":null,"role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":2,"collection":{"id":"e353f61379db4140","slug":"agile-software-development","title":"Agile Software Development","owner":{"id":"5bfcd1f236454eeb","login":"troymagennis","name":"Troy Magennis"}},"type":"notebook","body":{"titling":"auto","title":"How Many Samples Do I Need?","cells":[{"id":0,"value":"# How Many Samples Do I Need?\n\nWhen collecting data from a system or source you want to estimate the chance that the NEXT sample is within the range of what you have previously observed from that system. For example, as you finish work and collect cycle-time per item, or count the number of completed items per week (throughput per week), how many samples should you expect to need to be 80%, 90%, 95% certain for instance?\n\nIt turns our much less than you might expect. But there is one assumption that you need to consider. The samples should be random, and this is rarely true in software development because the next sample is just the next sample, but as long as you consider that work was started by some specific ordering criteria other than \"effort\" or \"size\" (something like value, or urgency\" you are pretty safe.)\n\n\"What is the likelihood the next sample is within the range of the lowest and highest sample seen so far\"\n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":17,"value":"md`${tex.block`\n\\% Chance In Range = \\frac{n-1}{n+1}\n`} where n = number of current samples.`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":57,"value":"md`Whenever we sample numbers, each sample can give new information. There are three positions the next ample could be. Lower than the lowest seen so far but equal to or greater than the lowest seen so far. Between the current lowest and lower than the current highest so far, or higher than the current seen or equal or less than the actual maximum. The formula shown computes the orange zone, and the remaining probability is shared by each orange zone.`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":67,"value":"viewof n = slider3({\n message: \"How many samples do you have so far? \",\n min: 2,\n max: 100,\n start: 2,\n step: 1,\n int: true\n})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":89,"value":"html` \n

actual lowest >

${(((1 - p) / 2) * 100).toLocaleString(undefined, {\n maximumFractionDigits: 1\n})}%

${(p * 100).toLocaleString(undefined, {\n maximumFractionDigits: 1\n})}%

${(((1 - p) / 2) * 100).toLocaleString(undefined, {\n maximumFractionDigits: 1\n})}%

< actual highest

actual lowest >

current lowest ^

current highest ^

\n`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":93,"value":"md`----\n### How does this work?\n\nWith two or more samples (n), there will be a current lowest and a current highest. If we divide the entire known space for samples to be between (an actual highest (unknown to us) and an actual lowest (also unknown to us) into equal possible segments, then we can compute the percentage of segments between the currently samples lowest and highest.\n\nFor example (the white dividers are sorted samples):`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":102,"value":"html` \n

when n = 2 samples

33%

when n = 3 samples

25%

when n = 4 samples

20%

`","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":146,"value":"$32","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":72,"value":"p = (n - 1) / (n + 1)","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":65,"value":"function slider3({\n message = \"Set your value\",\n min = 0,\n max = 1,\n start = 0.5,\n step = 0.1,\n precision = 3,\n int = false\n}) {\n min = min / step;\n max = max / step;\n start = start / step;\n\n const div = html`\n\t${message}\n \n ${start * step}\n `;\n const range = div.querySelector(\"[type=range]\");\n const value_disp = div.querySelector(\"span\");\n div.value = value_disp.innerHTML = start * step;\n\n range.addEventListener(\"input\", () => {\n const newValue = range.valueAsNumber * step;\n div.value = newValue;\n value_disp.innerHTML = int\n ? Math.round(newValue)\n : newValue.toPrecision(precision);\n });\n return div;\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W33","stdlib":"1"}} 31:"$30:body"