---
title: 'TimeSeriesScientist: Agentic Forecasting Framework'
url: https://www.emergentmind.com/topics/timeseriesscientist-tsci
type: topic
---

# TimeSeriesScientist: Agentic Forecasting Framework

TimeSeriesScientist, abbreviated **TSci**, is an LLM-driven agentic framework for general time series forecasting that treats forecasting as a staged decision process rather than as a single model-fitting step. It is designed for settings with thousands of short, noisy series that vary in frequency, quality, and forecast horizon, where the dominant cost lies in preprocessing, validation, and ensembling rather than in fitting an individual predictor. The framework consists of four specialized agents—**Curator**, **Planner**, **Forecaster**, and **Reporter**—and produces both forecasts and a comprehensive, transparent report with natural-language rationales [2510.01538].

## 1. Definition, scope, and problem setting

TSci addresses **univariate forecasting** from raw input data \(\mathcal{D}=\{x_t\}_{t=1}^T\). Its stated objective is a **general, domain-agnostic framework that minimizes human intervention** while remaining interpretable and extensible across tasks. In this formulation, the forecasting workflow is decomposed into diagnosis and preprocessing, model-pool selection and tuning, adaptive model selection and ensembling, and report synthesis [2510.01538].

A central design claim is that prevailing statistical and deep learning models are often tailored to specific datasets or domains and generalize poorly, whereas forecasters in practice must manage heterogeneous operational burdens before any final prediction is produced. TSci therefore positions the main unit of automation not as a single predictor but as an end-to-end workflow controller. A common misconception is to equate TSci with a new forecasting architecture analogous to a Transformer or a state-space model; the framework is instead a **tool-augmented multi-agent system** that orchestrates diagnostics, model selection, validation, and reporting [2510.01538].

Within the broader time-series software landscape, this places TSci alongside, but conceptually above, systems that solve narrower layers of the stack. The **tsibble** framework formalizes temporal data with an explicit index and key for tidy pipelines [1901.10257], **cesium** provides an end-to-end platform for featurization and machine learning with notebook export [1609.04504], and **GluonTS** supplies probabilistic forecasting abstractions such as Estimator, Predictor, Forecast, and Evaluator [1906.05264]. This suggests that TSci’s distinctive contribution is not a new base forecaster alone, but an agentic orchestration layer over the forecasting lifecycle.

## 2. Four-agent architecture and staged workflow

TSci organizes forecasting through four specialized agents: **Curator**, **Planner**, **Forecaster**, and **Reporter**. At a high level, the pipeline is described as follows: raw time series \(\mathcal{D}\) enter Curator; Curator produces cleaned data, visualizations, and analysis; Planner narrows the model hypothesis space and tunes candidates; Forecaster fits, validates, and selects a final model configuration or ensemble strategy; Reporter synthesizes the process into a human-readable report [2510.01538].

The workflow can be written as a sequence of transformations:
\[
(Q,V,A)\gets \mathrm{Curator}(\mathcal{D}), \qquad
\mathcal{M}_p\gets \mathrm{Planner.SelectModels}(Q,V,A),
\]
\[
(\mathcal{M}_{\mathrm{selected}},S_{\mathrm{val}})\gets \mathrm{Planner.Tune}(\mathcal{M}_p,\tilde{\mathcal{D}}_{\mathrm{val}}),
\]
\[
\hat y_{t+1:t+H}\gets \mathrm{Forecaster}(\mathcal{M}_{\mathrm{selected}},S_{\mathrm{val}}),
\qquad
\mathcal{R}\gets \mathrm{Reporter}(Q,V,A,\mathcal{M}_{\mathrm{selected}},\hat y).
\]

This decomposition is operational rather than merely descriptive. Curator outputs diagnostic artifacts, Planner outputs a selected subset of models and their tuned configurations, Forecaster outputs final predictions, and Reporter outputs a structured report. Because each stage exposes intermediate products, the framework is described as a **white-box system** with transparent natural-language rationales and comprehensive reports [2510.01538].

The architecture also makes explicit that TSci separates **representation of evidence** from **forecast generation**. Diagnostics \(Q\), visualizations \(V\), and temporal-structure analysis \(A\) are produced before model selection, rather than being implicit side products of a learned end-to-end network. This suggests an emphasis on procedural interpretability rather than solely post hoc explanation.

## 3. Curator and Planner

The **Curator** agent performs three steps: **Quality Diagnostics**, **Visualization Generation**, and **Temporal Structure Profiling**. It computes
\[
Q=\mathcal{A}_f(\mathcal{D})=(S,M,O,\pi),
\]
where \(S=(\mu,\sigma,x_{\min},x_{\max},\tau_{\mathrm{trend}})\), \(M\) and \(O\) record missing-value and outlier indices, and \(\pi=(m^*,h^*)\) is the chosen preprocessing strategy; it then applies a transformation \(\phi\) to produce cleaned data \(\tilde{\mathcal{D}}=\phi(\mathcal{D})\) [2510.01538].

Outlier detection in Curator includes three explicit schemes. Under **Rolling IQR**, \(x_t\) is an outlier if it falls below \(Q_1-1.5\,\mathrm{IQR}_t\) or above \(Q_3+1.5\,\mathrm{IQR}_t\) on a rolling window. Under **Rolling Z-Score**, \(x_t\) is an outlier if \(|z_t|>\alpha\) with \(3\le \alpha \le 4\). Under the **Percentile Rule**, an observation is flagged when it lies below \(P_{\mathrm{lower}}\) or above \(P_{\mathrm{upper}}\). The paper states that outlier handling includes clipping, interpolation, forward/backward fill, and smoothing [2510.01538].

For stationarity assessment, Curator may call external tools for the **Augmented Dickey–Fuller (ADF) test** and the **KPSS test**. It also generates three plot types through LLM-driven prompts with Matplotlib and seaborn: a time-series overview with rolling mean and standard deviation, a seasonal decomposition \(X_t=T_t+S_t+R_t\), and ACF/PACF plots up to lag 40. From cleaned data \(\tilde{\mathcal{D}}\) and visualizations \(V\), it prompts the LLM to extract
\[
A=(t,s,u),
\]
where \(t\) is trend type, \(s\) seasonality strength, and \(u\) the stationarity verdict [2510.01538].

The **Planner** then narrows the model hypothesis space in three stages: model selection, hyperparameter optimization, and model ranking. It maps multimodal diagnostics \((Q,V,A)\) to a candidate pool
\[
\mathcal{M}_p=\mathrm{Select}(\mathcal{M};n_p),
\]
where \(\mathcal{M}\) is a predefined library of **21 models**, including **ARIMA, ETS, Prophet, XGBoost, LSTM, Transformer, etc.** For each candidate \(m_i\in\mathcal{M}_p\), Planner samples \(\mathcal{C}_i\subset\Theta_i\) of size \(N\) and minimizes validation MAPE:
\[
\theta_i^*=\arg\min_{\theta\in\mathcal{C}_i}\mathrm{MAPE}_{\mathrm{val}}(m_i(\theta)).
\]
It then ranks tuned candidates by validation MAPE and keeps the top \(k\) models as
\[
\mathcal{M}_{\mathrm{selected}}=\{m_{(1)}(\theta_{(1)}^*),\dots,m_{(k)}(\theta_{(k)}^*)\}.
\]
This design explicitly couples diagnosis-driven model narrowing with metric-driven tuning rather than performing blind model sweep over the full library [2510.01538].

## 4. Forecaster and Reporter

The **Forecaster** agent takes the selected models \(\mathcal{M}_{\mathrm{selected}}\) and their validation metrics \(S_{\mathrm{val}}\), aggregates multiple metrics \(\ell_{i,m}\) into a scalar score,
\[
s_i=\sum_m \alpha_m\,\mathrm{norm}(\ell_{i,m}), \qquad \sum_m \alpha_m=1,
\]
and chooses an ensemble strategy [2510.01538].

Three ensemble strategies are specified. Under **Single-Best**, the chosen model is
\[
i^*=\arg\min_i s_i,\qquad w_{i^*}=1.
\]
Under **Performance-Aware Averaging**, the weights are defined through
\[
\tilde w_i=(s_i+\varepsilon)^{-\beta},\qquad
w_i\propto \tilde w_i^{1/\tau},
\]
followed by clipping and shrinkage toward uniform weighting. Under **Robust Aggregation**, TSci uses the **median** or a **trimmed mean** across model forecasts. Final evaluation reports **test MAE** and **RMSE** [2510.01538].

The **Reporter** agent synthesizes a comprehensive report \(\mathcal{R}\) with five sections:  
1. ensemble forecast \(\hat x^{\mathrm{ens}}_{t+1:t+H}\) with confidence intervals;  
2. performance summary for each model and the ensemble;  
3. interpretability report covering model selection, hyperparameter choices, ensemble weights, and forecast confidence and assumptions;  
4. visualization suite for preprocessing and forecast plots;  
5. workflow documentation as a decision-provenance log [2510.01538].

This reporting layer is a defining feature of TSci. Earlier systems expose reproducibility through data structures, notebooks, or evaluation APIs—such as tsibble’s explicit temporal metadata [1901.10257] or cesium’s JSON-logged notebook export [1609.04504]—but TSci makes report synthesis itself a first-class component of the forecasting system. This suggests that interpretability in TSci is procedural and documentary as much as statistical.

## 5. Experimental evaluation

TSci is evaluated on **eight public benchmarks covering five domains**. The datasets are listed with point counts and frequencies in the paper’s evaluation summary [2510.01538].

| Dataset(s) | Frequency | #points |
|---|---:|---:|
| ETTh1, ETTh2 | 1 h | 17,420 |
| ETTm1, ETTm2 | 15 m | 69,680 |
| Weather | 10 m | 52,696 |
| Electricity (ECL) | 1 h | 26,304 |
| Exchange | 1 d | 7,588 |
| ILI | 1 w | 966 |

The study reports **MAE** and **MAPE** and compares TSci against both **statistical baselines**—ARIMA, ETS, TBATS—and **LLM-based baselines**—GPT-4o, Gemini-2.5 Flash, Qwen-Plus, DeepSeek-v3, and Claude-3.7. The headline result is that **TSci reduces error by 10.4% versus statistical methods and 38.2% versus LLM baselines**, with full results by dataset and horizon reported in the appendix [2510.01538].

The benchmark mix is operationally significant. ETTh/ETTm and Weather provide higher-frequency settings, Exchange provides daily data, and ILI provides weekly data. This suggests that the framework is being evaluated across heterogeneous sampling regimes rather than only on a single industrial benchmark family.

## 6. Relation to adjacent methods, misconceptions, and limitations

TSci belongs to a broader line of systems that aim to reduce the engineering burden of time-series analysis, but its emphasis differs from feature-extraction platforms and probabilistic forecasting libraries. **hctsa** automates massive feature extraction over approximately 7,700 time-series features and then applies univariate or multivariate selection and modeling [1612.05296]. **cesium** provides featurization, model building, prediction, and reproducible notebook export through a Python library and web interface [1609.04504]. **GluonTS** standardizes probabilistic model development, training, prediction, and evaluation for forecasting workloads [1906.05264]. By contrast, TSci frames the entire workflow as an LLM-orchestrated sequence of diagnostic, selection, ensembling, and reporting decisions [2510.01538].

A second common misconception is that forecasting performance alone exhausts the contribution. The paper instead stresses **interpretability**, **transparency**, and **extensibility**: each agent logs clear rationales, the final report is fully transparent, and new diagnostics, models, or ensemble strategies can be plugged into the framework. This suggests that TSci should be understood as a forecasting operations framework as much as a benchmarked predictor [2510.01538].

The paper also states three principal limitations and future directions. First, **efficiency** remains constrained by LLM-call latency, making batching and caching open issues. Second, extending Curator’s diagnostics to **multivariate or multimodal series** is identified as a challenge. Third, the framework leaves open **knowledge infusion**, including integration of exogenous signals or external knowledge bases [2510.01538].

These limitations delineate the current scope of TSci. It is presented as a general-purpose AI agent for forecasting, but its published formulation centers on univariate series and on workflow automation around a predefined model library. Its contribution therefore lies in turning preprocessing, model selection, validation, ensembling, and reporting into an explicit, inspectable agent pipeline rather than in replacing the underlying forecasting methods with a single new architecture [2510.01538].

Source: https://www.emergentmind.com/topics/timeseriesscientist-tsci