TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Prophet: forecasting at scale

520 点作者 benhamner大约 8 年前

33 条评论

confounded大约 8 年前
Worth noting Prophet is R&#x2F;Python wrappers to some models with reasonable defaults, written in and fit by <i>Stan</i>, a probabilistic programming language, and Bayesian estimation framework.<p>Stan is amazing in that you can fit pretty much any model you can describe in an equation (given enough time and compute, of course)!<p>More on Stan here: <a href="http:&#x2F;&#x2F;mc-stan.org&#x2F;" rel="nofollow">http:&#x2F;&#x2F;mc-stan.org&#x2F;</a>
评论 #13743071 未加载
评论 #13743454 未加载
评论 #13743727 未加载
rodionos大约 8 年前
I didn&#x27;t know wikipedia page view counters are available for public usage.<p>The wikipediatrend R package relies on <a href="http:&#x2F;&#x2F;stats.grok.se&#x2F;" rel="nofollow">http:&#x2F;&#x2F;stats.grok.se&#x2F;</a>, which in turn relies on <a href="https:&#x2F;&#x2F;dumps.wikimedia.org&#x2F;other&#x2F;pagecounts-raw&#x2F;" rel="nofollow">https:&#x2F;&#x2F;dumps.wikimedia.org&#x2F;other&#x2F;pagecounts-raw&#x2F;</a> which has been deprecated.<p>The new dump is located at <a href="https:&#x2F;&#x2F;dumps.wikimedia.org&#x2F;other&#x2F;pageviews&#x2F;" rel="nofollow">https:&#x2F;&#x2F;dumps.wikimedia.org&#x2F;other&#x2F;pageviews&#x2F;</a><p>Data is available in hourly intervals.<p>* pageviews-20170227-050000<p><pre><code> en Peyton_Manning 58 0 </code></pre> [edit] There is a wikipedia-hosted OSS viewer for these logs, e.g. Swedish crime stats:<p><a href="https:&#x2F;&#x2F;tools.wmflabs.org&#x2F;pageviews&#x2F;?project=en.wikipedia.org&amp;platform=all-access&amp;agent=user&amp;range=latest-90&amp;pages=Crime_in_Sweden" rel="nofollow">https:&#x2F;&#x2F;tools.wmflabs.org&#x2F;pageviews&#x2F;?project=en.wikipedia.or...</a>
评论 #13743332 未加载
评论 #13743489 未加载
评论 #13743330 未加载
评论 #13742982 未加载
saosebastiao大约 8 年前
This is an interesting project, and in one of the areas where almost all businesses could do better. Anecdotally, there is a ton of money left on the table by established businesses that do it poorly, which also leaves lots of room for resume-padding technical experience. So anything that claims to improve the state of the art of automated forecasting is definitely worth watching.<p>That being said this claim in point #1 baffles me:<p>&gt; Prophet makes it much more straightforward to create a reasonable, accurate forecast. The forecast package includes many different forecasting techniques (ARIMA, exponential smoothing, etc), each with their own strengths, weaknesses, and tuning parameters. We have found that choosing the wrong model or parameters can often yield poor results, and it is unlikely that even experienced analysts can choose the correct model and parameters efficiently given this array of choices.<p>The forecast package contains an auto.arima function which does full parameter optimization using AIC which is just as hands free as is claimed of Prophet. I have been using it commercially and successfully for years now. Maybe prophet produces better models (I&#x27;ll definitely take a look myself), but to claim that it&#x27;s not possible to get good results without experience seems a bit disingenuous.<p>As an aside, anybody interested in a great introductory book on time series forecasting should check out Rob Hyndman&#x27;s book which is freely available online. <a href="https:&#x2F;&#x2F;www.otexts.org&#x2F;fpp" rel="nofollow">https:&#x2F;&#x2F;www.otexts.org&#x2F;fpp</a>
评论 #13747481 未加载
schlarpc大约 8 年前
Moderately relevant short story: <a href="https:&#x2F;&#x2F;www.facebook.com&#x2F;notes&#x2F;robin-sloan&#x2F;julie-rubicon&#x2F;985697811525170" rel="nofollow">https:&#x2F;&#x2F;www.facebook.com&#x2F;notes&#x2F;robin-sloan&#x2F;julie-rubicon&#x2F;985...</a>
评论 #13742594 未加载
techno_modus大约 8 年前
It seems that they have developed a model for only univariate forecasts and only numeric regular time series which is a classical use case in statistics. Yet, most data sources have many dimensions (for example, energy consumption, temperature, humidity etc.) as well as categorical data like current state (On, Off). The situation is even more difficult if the data is not a regular time series but is more like asynchronous event stream. It would be interesting to find a good forecasting model for some of these use cases. In particular, it is interesting if this Prophet model can be generalized and applied to multivariate data.
评论 #13744343 未加载
cardosof大约 8 年前
That&#x27;s very cool, congrats and thank you to the Facebook guys!<p>A few days ago I was asked to do some forecasting with a daily revenue series for a client. Due to her business&#x27; nature the series was really tricky with weekdays and months&#x2F;semesters having some specific effects on the data. I as many use Hyndman&#x27;s forecast package, but I threw this data at prophet and it delivered a nice plot with the (correct) overall trend and seasonalities. Very cool and easy to do something.
yoghurtio大约 8 年前
We at <a href="https:&#x2F;&#x2F;yoghurt.io&#x2F;" rel="nofollow">https:&#x2F;&#x2F;yoghurt.io&#x2F;</a> have been working towards similar forecasting solution. So far the feedback has been that automated solutions can also bring good results at a far lesser cost compared to hiring an expert analyst.<p>Its a completely managed solution. No need to setup anything yourself.Just upload the data and predict next week&#x27;s data, today itself. There is a free trial and if anyone here is looking for an extended trial, they can reach out to me.
评论 #13742646 未加载
anacleto大约 8 年前
This is so great!<p>I&#x27;ve been using CasualImpact by Google [0] for months. This seems pretty straightforward.<p>[0] <a href="https:&#x2F;&#x2F;google.github.io&#x2F;CausalImpact&#x2F;CausalImpact.html" rel="nofollow">https:&#x2F;&#x2F;google.github.io&#x2F;CausalImpact&#x2F;CausalImpact.html</a>
jl6大约 8 年前
I wonder what Sungard&#x2F;FIS think of the name, which is the same as their commercial financial modelling&#x2F;forecasting tool.
评论 #13743782 未加载
pacifika大约 8 年前
The more facebook grows the more tools it aligns tooling with intelligence services.
asafira大约 8 年前
So...How much will this do at forecasting stock prices? =)<p>Very cool though --- I would be interested to dive into the methods they&#x27;ve implemented sometime in the near future!
评论 #13742296 未加载
评论 #13745394 未加载
评论 #13742281 未加载
hnarayanan大约 8 年前
Is there a way to extend these models to handle spatial variation (e.g. weather forecasting, property price estimation etc.) as well?
评论 #13742926 未加载
dmichulke大约 8 年前
I have been working for a few years on a similar project using evolutionary algorithms on top of other models (linear &#x2F; ann). It works quite well (e.g., for equidistant energy demand &#x2F; supply forecasts) but there&#x27;s still lots of stuff to do.<p>It&#x27;s major benefit is that it figures out relationship to the target time series by itself, so you can just throw in all time series and see what comes out.<p>Language is Clojure, 20kloc, incanter, encog. If anyone is interested in working for&#x2F;with it, let me know. I currently develop a Rest Api for it and plan to release it as open source once the major code smells are dealt with.
评论 #13744083 未加载
Steeeve大约 8 年前
This actually looks incredibly useful and pretty simple to learn.<p>Between this and Stan I think my free time for the next week is gone.
zebrafish大约 8 年前
So.... I don&#x27;t understand how this is better or worse than using forecast.<p>You talk about having to choose the best algorithm but it seems like Prophet is just another algorithm to choose from. Is there some kind of built in grid-search or are you just stating that results from your AM have been more accurate than ARIMA?
hn_username大约 8 年前
This is a nice piece of work - thanks for sharing with the community!<p>Some feedback: it&#x27;d be nice to see you actually quantify how accurate Prophet&#x27;s forecasts are on the landing page for the project. In the Wikipedia page view example, you go as far as showing a Prophet forecast, but it&#x27;d be nice to have you take it one step further and quantify its performance. Maybe withhold some of the data you use to fit the model and see how it performs on that out of sample data. It&#x27;s nice that you show qualitatively that it captures seasonality, but you make bold claims about its accuracy and the data to back those claims up is conspicuously absent. Related, it might be worth benchmarking its performance against existing automated forecasting tools.<p>I&#x27;ll definitely be checking it out!
SmellTheGlove大约 8 年前
For us insurance&#x2F;financial services folks, I would like to simply clarify that this is not the Sungard&#x2F;FIS risk management platform that is also called Prophet! :D<p>I got really excited for a second. Actually, I&#x27;m still pretty excited about this even if it was something else entirely.
nickfzx大约 8 年前
This looks amazing, congratulations.<p>We&#x27;re planning to add forecasting to our SaaS analytics product (<a href="https:&#x2F;&#x2F;chartmogul.com" rel="nofollow">https:&#x2F;&#x2F;chartmogul.com</a>) later this year, I&#x27;m going to look and see if we can use this in our product now.
评论 #13745018 未加载
minimaxir大约 8 年前
Interesting definition of &quot;scale&quot; in this context, as it does <i>not</i> imply &quot;big data&quot; like every other usage of the word scale in data science. The tool works on, and is optimized, for day-to-day, mundane data.<p>See also the R vignette, which shows that the data is returned per-column which gives it a lot of flexibility if you only want certain values: <a href="https:&#x2F;&#x2F;cran.r-project.org&#x2F;web&#x2F;packages&#x2F;prophet&#x2F;vignettes&#x2F;quick_start.html" rel="nofollow">https:&#x2F;&#x2F;cran.r-project.org&#x2F;web&#x2F;packages&#x2F;prophet&#x2F;vignettes&#x2F;qu...</a>
paulvs大约 8 年前
For a corporate credit analyst working at a bank, what are some good introduction material for getting into forecasting using tools like these?<p>I see this being applicable to analysts when deciding on on a company&#x27;s credit worthiness.
评论 #13745173 未加载
syntaxing大约 8 年前
The fact that Prophet follows the &quot;sklearn model API&quot; and that it&#x27;s very well integrated with pandas makes it super appealing and usable!
monkeydust大约 8 年前
Very cool, got loads of sensor data around my house over a years worth so curious to throw it at Prophet.<p>Has anyone managed to get this working on windows with Juypter (Anaconda build) struggling with Pystan errors. Any guidance welcomed.
eternalban大约 8 年前
&#x2F;please ignore: Oracle &amp; Prophet. Oracle sifts through signs but Prophet has a line to the larger picture. I suppose the next &#x27;product&#x27; will be called Messiah to complete the picture.
elwell大约 8 年前
Why do we need Prophet when we already have Temple OS (<a href="http:&#x2F;&#x2F;www.templeos.org&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.templeos.org&#x2F;</a>)?
nodesocket大约 8 年前
Are there any startups&#x2F;services where you pass it a series and it returns forecast models? That&#x27;s something I&#x27;d be willing to pay for.
评论 #13742852 未加载
alexpetralia大约 8 年前
Slightly inconvenient that the main image &lt;figure&gt; needs to be replaced by an &lt;img&gt; tag just to have the image appear in print outs.
poppingtonic大约 8 年前
This is very interesting. Forecasters who participate in the Good Judgment Project, such as myself, will find this useful.
recurser大约 8 年前
Very cool. Could this be re-purposed for detecting anomalies&#x2F;outliers in time series data?
评论 #13743168 未加载
评论 #13743138 未加载
ayayecocojambo大约 8 年前
Can we use other features (like temperatue?), or it has to be only time-based?
agounaris大约 8 年前
How different this framework is from statsmodels?
评论 #13742469 未加载
hubot大约 8 年前
can someone explain what&#x27;s the meaning of this line<p>&gt; df[&#x27;y&#x27;] = np.log(df[&#x27;y&#x27;])
评论 #13744677 未加载
评论 #13744598 未加载
Helmet大约 8 年前
just wanted to point out to potential windows users - this will only run on python 3.5 due to dependencies (pystan only works on python 3.5 for windows)
fagnerbrack大约 8 年前
Facebook...