Canonical: https://instain.ru/en/ai/amazon-chronos/ Language: en Amazon Science the Time Series Forecast Model # Amazon Chronos. Open Amazon Science models for numerical time series 4.30 / 5 [Directory score](https://instain.ru/en/ai/amazon-chronos/#editorial-rating) [Official website ](https://github.com/amazon-science/chronos-forecasting) The Amazon Chronos family is designed to predict time series. Variants of the family differ in the support of joint series and additional features. ![Amazon Science Repository with Chronos Model Description](https://instain.ru/assets/images/amazon-chronos-screen-aa0b421198c4-1200.webp) Amazon Science Repository with Chronos models description. [Open the source ](https://github.com/amazon-science/chronos-forecasting) ## Forecast with a mark on the moment of launch Weekly demand may look predictable until a sale or a stock shortage. Amazon Chronos forecasts from historical time series. For a practice test, start at a date when you could actually have made the forecast. Record which observations were available then. A holiday calendar known in advance can be an extra feature in a compatible variant. Do not give the model sales figures from the forecast week. ## Comparison of several segments Choose several consecutive evaluation dates. For each date, use the earlier history to forecast the next week and compare it with observed sales. Add a simple benchmark, such as sales on the same weekday a week earlier. If the model returns intervals, check their width and how often observations fall inside them. Examine promotion days and rare spikes separately. An average error can conceal weaknesses that matter for purchasing. ## Where the model ends Missing a date and selling zero mean different things. Record how both situations are handled before comparing with a landmark. When changing the warehouse, assortment or accounting rules, repeat the check on the new history site. The forecast provides a numeric basis for discussion, and stock decisions require taking into account the price of the error and the actual supply constraints. Before the next cycle, save the original unloading and the rule for choosing control dates. ## Suitable tasks - Forecast weekly demand by sales history - Compare forecast intervals with a simple seasonal benchmark Features ## Strengths - Chronos maintains forecast with uncertainty estimate - Developer publishes code and examples of series preparation Things to consider ## Limitations - Only use information that was available at the time of the forecast - Accuracy is not guaranteed when demand patterns change ## Selection and conditions Free access Yes Trial period Not confirmed No sign-up required Not confirmed Russian interface Not confirmed Local installation Yes API Not confirmed “Not confirmed” means there is not enough information. The feature may still be available; check the developer’s terms. Input formats Data Result Data ### Terms Model weights are available under the terms in their model cards. You supply compute for local runs. Managed compute is billed under the platform’s terms. Only use information that was available at the time of the forecast. From task to first result ## Practical example Task Forecast weekly demand by sales history What to prepare The training range of daily sales and the calendar of holidays, known on the forecast date. ### Task brief or prompt Make a weekly forecast for several past dates using only the information available for each date. ### Expected result A table of forecasts and facts with error, intervals and a simple seasonal benchmark. ### How to check the result - Check the absence of future data at the entrance. - Separately compare the usual weeks and periods of shares. An editorial teaching example of a task brief, not the result of hands-on testing of the service. Check that the required mode is available, that you have rights to the source material and that the intended use of the result is permitted. Check details ## Official sources - [Official Amazon Chronos repository ](https://github.com/amazon-science/chronos-forecasting) - [Chronos 2 Card ](https://huggingface.co/amazon/chronos-2) This description is based on the developer’s material. Features, prices and availability may change; check the primary source before use. About the tool Developer Amazon Science Category [Code](https://instain.ru/en/category/code/) Access Choose the official weights and a compatible library. Verify the time format, supported features, and memory requirements for your model variant. Payment Model weights are available under the terms in their model cards. You supply compute for local runs. Managed compute is billed under the platform’s terms. API No separate API specified [Developer terms ](https://huggingface.co/amazon/chronos-2) Directory score 4.30 / 5 Features 4.50 / 5 Workflows 4.00 / 5 Ecosystem 4.30 / 5 Documentation 4.40 / 5 Scores are an editorial guide to the directory. They are not user reviews or the results of personal testing. [See the methodology ](https://instain.ru/en/methodology/) Keep exploring ## For similar tasks [Full category ](https://instain.ru/en/category/code/) Code ### [TimesFM](https://instain.ru/en/ai/google-timesfm/) Forecasting from time series history with backtesting 4.38 / 5 the Time Series Forecast Model Similar tasks; compare terms, formats and limitations