An automated manufacturing unit uses 25 sensors to record data and these data are then used to predict the probable state of the unit. The sensors record the data in every minute and in every 10 minutes the state of the unit is also recorded. The state could be ‘no risk’, ‘low risk’, ‘medium risk’ and ‘catastrophic’. The goal is to build model that will take the data from 25 sensors and predict the state of the unit in every 10 minutes.

Training data format: The traning data is given in a file called “train.txt”. The file is comma separated. The first and the second field in each line give the sample id and the timepoint respectively. The other fields in the line are the normalized values from the sensors. After every 10 line, you have a line with only one integer, which is the state label of the unit. The integer value 0, 1, 2, 3 refer to no risk’, ‘low risk’, ‘medium risk’ and ‘catastrophic’ respectively.

Test data: The file named ‘test.txt’ contains the test data. Once again, the data has the same format as the train data but without the true labels.

This dataset was created to investigate the impact of data collection modes and pre-processing techniques on the quality of free comment data related to consumers' sensory perceptions. A total of 200 consumers were recruited and divided into two groups of 100. Each group evaluated six madeleine samples (five distinct samples and one replicate) in a controlled sensory analysis laboratory, using different free comment data collection modes. Consumers in the first group provided only words or short expressions, while those in the second group used complete sentences. Additionally, participants reported their liking for each sample.

Automated in situ soil sensor network - the data set includes hourly and daily measurements of volumetric water content, soil temperature, and bulk electrical conductivity, collected at 42 monitoring locations and 5 depths (30, 60, 90, 120, and 150 cm) across Cook Agronomy Farm. Data collection was initiated in April 2007 and is ongoing. Description of data Tabular data CAF_sensors: folder with Daily and Hourly subfolders, each containing 42 '.txt' files of water content and temperature sensor readings. Each file represents readings from a single location, indicated in the file name (i.e. CAF003.txt) and in the 'Location' field of the table. Readings are organized by 'Date' (4/20/2007 - 6/16/2016), ‘Time’ (24 hr clock, only in hourly files), and with property (VW or T) and sensor 'Depth' as follows: VW_30cm: volumetric water readings at 30 cm depth (m^3/m^3) VW_60cm: volumetric water readings at 60 cm depth (m^3/m^3) VW_90cm: volumetric water readings at 90 cm depth (m^3/m^3) VW_120cm: volumetric water readings at 120 cm depth (m^3/m^3) VW_150cm: volumetric water readings at 150 cm depth (m^3/m^3) T_30cm: temperature readings at 30 cm depth (C) T_60cm: temperature readings at 60 cm depth (C) T_90cm: temperature readings at 90 cm depth (C) T_120cm: temperature readings at 120 cm depth (C) T_150cm: temperature readings at 150 cm depth (C) Volumetric water content readings are calibrated according to: Gasch, CK, DJ Brown, ES Brooks, M Yourek, M Poggio, DR Cobos, CS Campbell. 2017. A pragmatic, automated approach for retroactive calibration of soil moisture sensors using a two-step, soil specific correction. Computers and Electronics in Agriculture, 137: 29-40. Temperature readings are factory calibrated. CAF_BulkDensity.txt: file containing bulk density values ('BulkDensity' in g/cm^3) for sensor depths at each of the 42 instrumented locations at Cook Farm. Location is indicated in 'Location' field, and sample depths are defined (in cm) by the ’Depth’ field. CAF_CropID.txt: file containing crop codes for each sub-field (A, B and C) and strip (1-6 for A and B, 1-8 for C) at Cook Farm for 2007-2016. This is also part of the attribute table for 'CAF_strips.shp' CAF_CropCodes.txt: file containing crop code names and crop identities, used in 'CAF_CropID.txt' and 'CAF_strips.shp' CAF_ParticleSize.txt: file containing particle size fractions ('Sand', 'Silt', and 'Clay' as percent) for each 'Location' at sensor depths ('Depth', in cm). Spatial data All spatial data have spatial reference NAD83, UTM11N CAF_sensors.shp: file containing locations of each of the 42 monitoring locations, the 'Location' field contains the location name, which coincides with locations in tabular files. CAF_strips.shp: file containing areal extents of each sub-field, stip, and crop identities for 2007-2016. Crop identity codes are listed in 'CAF_CropCodes.txt' CAF_DEM.tif: file containing a 10 x 10 m elevation (in m) grid for Cook Farm. CAF_Spring_ECa.tif, CAF_Fall_ECa.tif: files containing 10 x 10 m apparent electrical conductivity (dS/m) grids to 1.5 m depth for spring and fall at Cook Farm. CAF_Bt_30cm.tif, CAF_Bt_60cm.tif, CAF_Bt_90cm.tif, CAF_Bt_120cm.tif, CAF_Bt_150cm.tif: files containing 10 x 10 m predictive surfaces for probability (0-1) of Bt horizon at the five sensor depths. Quality Control The Flags folder consists of the files containing the quality control flags for the Cook Farm Sensor Dataset. The nomenclature for the files indicates flags for either temperature (T) or water content (VW) and sensor depths. For example: T_30 is for the temperature data at 30cm. depth VW_120 is for the Volumetric water content at 120 cm. depth Files starting with “missing” contain flags (“M”) for locations and dates (mm/dd/yyyy) with missing data (NA in original dataset). Files starting with “range” contain flags for locations and dates (mm/dd/yyyy) with values outside acceptable ranges: Soil moisture (0-0.6 m^3/m^3) flagged as “C” Soil temperature (<0 deg. C) flagged as “D” Files starting with the name “flats” contain flags (“D”) for locations, dates (mm/dd/yyyy), and times (hh:mm) with constant values (within 1%) for a 24 hour period, as in Dorigo et al. 2013. Files starting with the name “spikes” contain flags (“D”) for locations, dates (mm/dd/yyy), and times (hh:mm) with sudden spikes in VWC readings. Files starting with the name “breaks” contain flags (“D”) for locations, dates (mm/dd/yyy), and times (hh:mm) with sudden breaks (jumps or drops) in VWC readings. Code (implemented in R) for the screening and flagging is included in “Code Snippet.txt” A list of the sensor versions as of 06/16/16 at each location and depth. Resources in this dataset:Resource Title: Data package for automated in situ soil sensor network. File Name: CAF_Sensor_Dataset.zipResource Description: Data file descriptions for Cook Farm sensor network data set (CAF_Sensor_Dataset). Data set compiled by Caley Gasch, under supervision of David Brown, Department of Crop and Soil Sciences, Washington State University, Pullman, WA. Updated: 04/01/2017 Tabular data: CAF_sensors: folder with Daily and Hourly subfolders, each containing 42 '.txt' files of water content and temperature sensor readings. Each file represents readings from a single location, indicated in the file name (i.e. CAF003.txt) and in the 'Location' field of the table. Readings are organized by 'Date' (4/20/2007 - 6/16/2016), ‘Time’ (24 hr clock, only in hourly files), and with property (VW or T) and sensor 'Depth' as follows: VW_30cm: volumetric water readings at 30 cm depth (m^3/m^3) VW_60cm: volumetric water readings at 60 cm depth (m^3/m^3) VW_90cm: volumetric water readings at 90 cm depth (m^3/m^3) VW_120cm: volumetric water readings at 120 cm depth (m^3/m^3) VW_150cm: volumetric water readings at 150 cm depth (m^3/m^3) T_30cm: temperature readings at 30 cm depth (C) T_60cm: temperature readings at 60 cm depth (C) T_90cm: temperature readings at 90 cm depth (C) T_120cm: temperature readings at 120 cm depth (C) T_150cm: temperature readings at 150 cm depth (C) Volumetric water content readings are calibrated according to: Gasch, CK, DJ Brown, ES Brooks, M Yourek, M Poggio, DR Cobos, CS Campbell. 2017. A pragmatic, automated approach for retroactive calibration of soil moisture sensors using a two-step, soil specific correction. Computers and Electronics in Agriculture, 137: 29-40. Temperature readings are factory calibrated. CAF_BulkDensity.txt: file containing bulk density values ('BulkDensity' in g/cm^3) for sensor depths at each of the 42 instrumented locations at Cook Farm. Location is indicated in 'Location' field, and sample depths are defined (in cm) by the ’Depth’ field. CAF_CropID.txt: file containing crop codes for each sub-field (A, B and C) and strip (1-6 for A and B, 1-8 for C) at Cook Farm for 2007-2016. This is also part of the attribute table for 'CAF_strips.shp' CAF_CropCodes.txt: file containing crop code names and crop identities, used in 'CAF_CropID.txt' and 'CAF_strips.shp' CAF_ParticleSize.txt: file containing particle size fractions ('Sand', 'Silt', and 'Clay' as percent) for each 'Location' at sensor depths ('Depth', in cm). Spatial data: all spatial data have spatial reference NAD83, UTM11N CAF_sensors.shp: file containing locations of each of the 42 monitoring locations, the 'Location' field contains the location name, which coincides with locations in tabular files. CAF_strips.shp: file containing areal extents of each sub-field, stip, and crop identities for 2007-2016. Crop identity codes are listed in 'CAF_CropCodes.txt' CAF_DEM.tif: file containing a 10 x 10 m elevation (in m) grid for Cook Farm. CAF_Spring_ECa.tif, CAF_Fall_ECa.tif: files containing 10 x 10 m apparent electrical conductivity (dS/m) grids to 1.5 m depth for spring and fall at Cook Farm. CAF_Bt_30cm.tif, CAF_Bt_60cm.tif, CAF_Bt_90cm.tif, CAF_Bt_120cm.tif, CAF_Bt_150cm.tif: files containing 10 x 10 m predictive surfaces for probability (0-1) of Bt horizon at the five sensor depths. (Dataset updated on 10/23/2017 to include QC information.)

The synthetic sensor dataset contains more than 3000 samples, each representing a set of sensor readings. It consists of six columns: Temperature, Sensor1, Sensor2, Sensor3, Sensor4, and Sensor5.

The dataset is designed to mimic a scenario where temperature readings are influenced by multiple independent sensor measurements. The values of the independent variables and the added noise introduce variability in the temperature readings.

This resource contains Jupyter Notebooks with examples for conducting quality control post processing for in situ aquatic sensor data. The code uses the Python pyhydroqc package. The resource is part of set of materials for hydroinformatics and water data science instruction. Complete learning module materials are found in HydroLearn: Jones, A.S., Horsburgh, J.S., Bastidas Pacheco, C.J. (2022). Hydroinformatics and Water Data Science. HydroLearn. https://edx.hydrolearn.org/courses/course-v1:USU+CEE6110+2022/about.

This resources consists of 3 example notebooks and associated data files.

Notebooks: 1. Example 1: Import and plot data 2. Example 2: Perform rules-based quality control 3. Example 3: Perform model-based quality control (ARIMA)

Data files: Data files are available for 6 aquatic sites in the Logan River Observatory. Each file contains data for one site for a single year. Each file corresponds to a single year of data. The files are named according to monitoring site (FranklinBasin, TonyGrove, WaterLab, MainStreet, Mendon, BlackSmithFork) and year. The files were sourced by querying the Logan River Observatory relational database, and equivalent data could be obtained from the LRO website or on HydroShare. Additional information on sites, variables, and methods can be found on the LRO website (http://lrodata.usu.edu/tsa/) or HydroShare (https://www.hydroshare.org/search/?q=logan%20river%20observatory). Each file has the same structure indexed with a datetime column (mountain standard time) with three columns corresponding to each variable. Variable abbreviations and units are: - temp: water temperature, degrees C - cond: specific conductance, μS/cm - ph: pH, standard units - do: dissolved oxygen, mg/L - turb: turbidity, NTU - stage: stage height, cm

For each variable, there are 3 columns: - Raw data value measured by the sensor (column header is the variable abbreviation). - Technician quality controlled (corrected) value (column header is the variable abbreviation appended with '_cor'). - Technician labels/qualifiers (column header is the variable abbreviation appended with '_qual').

Sensory evaluation of traditional and pipe distillation of areke samples

This dataset provides a comprehensive sensory and pricing analysis of 116 commercial New Zealand Pinot Noir wines from Central Otago, Marlborough, and Martinborough. The dataset includes measurements of quality parameters, detailed mouthfeel and aroma attributes assessed by a trained sensory panel. Additionally, the dataset includes the actual retail prices of the wines alongside the panellists’ expected prices, offering insights into the perceived quality and valuation of wines. This data supports comparative studies on Pinot Noir from other regions, exploration of wine sensory quality and further investigation into wine regionality. Its structured approach facilitates replication, enhancing the understanding of Pinot Noir’s sensory and market profiles in relation to terroir.

This resource contains an example script for using the software package pyhydroqc. pyhydroqc was developed to identify and correct anomalous values in time series data collected by in situ aquatic sensors. For more information, see the code repository: https://github.com/AmberSJones/pyhydroqc and the documentation: https://ambersjones.github.io/pyhydroqc/. The package may be installed from the Python Package Index.

This script applies the functions to data from a single site in the Logan River Observatory, which is included in the repository. The data collected in the Logan River Observatory are sourced at http://lrodata.usu.edu/tsa/ or on HydroShare: https://www.hydroshare.org/search/?q=logan%20river%20observatory.

Anomaly detection methods include ARIMA (AutoRegressive Integrated Moving Average) and LSTM (Long Short Term Memory). These are time series regression methods that detect anomalies by comparing model estimates to sensor observations and labeling points as anomalous when they exceed a threshold. There are multiple possible approaches for applying LSTM for anomaly detection/correction. - Vanilla LSTM: uses past values of a single variable to estimate the next value of that variable. - Multivariate Vanilla LSTM: uses past values of multiple variables to estimate the next value for all variables. - Bidirectional LSTM: uses past and future values of a single variable to estimate a value for that variable at the time step of interest. - Multivariate Bidirectional LSTM: uses past and future values of multiple variables to estimate a value for all variables at the time step of interest.

The correction approach uses piecewise ARIMA models. Each group of consecutive anomalous points is considered as a unit to be corrected. Separate ARIMA models are developed for valid points preceding and following the anomalous group. Model estimates are blended to achieve a correction.

The anomaly detection and correction workflow involves the following steps: 1. Retrieving data 2. Applying rules-based detection to screen data and apply initial corrections 3. Identifying and correcting sensor drift and calibration (if applicable) 4. Developing a model (i.e., ARIMA or LSTM) 5. Applying model to make time series predictions 6. Determining a threshold and detecting anomalies by comparing sensor observations to modeled results 7. Widening the window over which an anomaly is identified 8. Aggregating detections resulting from multiple models 9. Making corrections for anomalous events

Instructions to run the notebook through the CUAHSI JupyterHub: 1. Click "Open with..." at the top of the resource and select the CUAHSI JupyterHub. You may need to sign into CUAHSI JupyterHub using your HydroShare credentials. 2. Select 'Python 3.8 - Scientific' as the server and click Start. 2. From your JupyterHub directory, click on the ExampleNotebook.ipynb file. 3. Execute each cell in the code by clicking the Run button.

The .cvs files in this research item contain the ground temperatures of the sample points. The latter are determined by defining a grid of equally spaced lines (50 mm distance between each other) on each coordinate axis.

File names follow the convention "TS_DATE.csv" where DATE can be either be 30-08-2017 or 31-08-2017 and TS stands for Thermal Sensor.

Pulse sample data

Sensory Analysis Software for Beverages Market Outlook

The global sensory analysis software for beverages market size reached USD 412.5 million in 2024, as per our latest research, and is expected to grow at a robust CAGR of 10.7% from 2025 to 2033. By the end of the forecast period in 2033, the market is projected to achieve a value of USD 1,028.6 million. This significant growth is primarily driven by the increasing demand for quality assurance, product innovation, and regulatory compliance in the beverage industry, propelled by the integration of advanced digital technologies and data analytics into sensory evaluation processes.

One of the primary growth factors fueling the expansion of the sensory analysis software for beverages market is the rising emphasis on product quality and consistency among beverage manufacturers. As global competition intensifies, brands are under mounting pressure to deliver beverages that consistently meet consumer expectations for taste, aroma, and mouthfeel. Sensory analysis software enables manufacturers to digitize and standardize sensory testing, ensuring objective, repeatable, and data-driven results. This not only streamlines quality control but also minimizes human error and subjectivity, leading to more reliable product releases. The adoption of such software is further accelerated by the increasing complexity of beverage portfolios, including the introduction of novel flavors, functional ingredients, and health-oriented products that demand rigorous sensory evaluation.

Another key driver is the growing importance of consumer-centric product development in the beverages sector. As consumer preferences evolve rapidly, beverage companies are leveraging sensory analysis software to gather, analyze, and interpret large volumes of sensory data from consumer panels and focus groups. This allows for agile product development cycles, enabling brands to fine-tune formulations and launch products that resonate with target demographics. The integration of artificial intelligence and machine learning into sensory analysis platforms further enhances predictive capabilities, allowing manufacturers to anticipate market trends and optimize product attributes for maximum consumer appeal. This trend is particularly pronounced in highly competitive segments such as craft beverages, plant-based drinks, and functional beverages, where differentiation hinges on nuanced sensory profiles.

Regulatory compliance and traceability are also significant growth catalysts for the sensory analysis software for beverages market. As food safety regulations become more stringent worldwide, beverage producers are required to maintain comprehensive documentation of sensory testing protocols and outcomes. Sensory analysis software facilitates seamless compliance by automating data capture, audit trails, and reporting, thereby reducing the risk of non-compliance and associated penalties. Additionally, the software supports cross-functional collaboration between R&D, quality assurance, and regulatory teams, fostering a culture of transparency and continuous improvement. The increasing adoption of digital transformation initiatives across the beverage industry further amplifies the demand for integrated sensory analysis solutions.

From a regional perspective, North America and Europe currently dominate the sensory analysis software for beverages market, accounting for a substantial share of global revenue. This leadership is attributed to the presence of major beverage manufacturers, advanced research infrastructure, and a strong focus on quality standards. However, the Asia Pacific region is emerging as the fastest-growing market, driven by rapid industrialization, rising disposable incomes, and the expansion of the beverage sector in countries such as China, India, and Japan. Latin America and the Middle East & Africa are also witnessing steady growth, supported by increasing investments in food and beverage processing technologies and a growing emphasis on export-oriented production. Regional dynamics are further shaped by differences in regulatory frameworks, consumer preferences, and technological adoption rates, making localized strategies essential for market participants.

Eight different datasets using Polarized Sensory Positioning (PSP) for evaluation of food/drink samples.

The different datasets are given in separate sheets of the file. The datasets are further described in sheet "INFO" and samples for different datasets are described with the data.

Useful for comparison of data analytical methods.

The ranking results of all assessor participating in the experiment. The data presented present the results of sensory tests in the field of checking the sensory sensitivity of the sense of touch. the data is used to compare the assessors' efficiency in the ability to sort samples using the sense of touch. Set 1 Sandpaper Set 2 Thickness of a sheet of paper Set 3 PVC cones with plasticizer additive Set 4 Model emulsions

📄 Dataset Description The FlavorSense dataset captures human responses to various flavor attributes, aiming to model and predict taste preferences using machine learning. It includes both categorical and numerical features such as sweetness, bitterness, aroma, aftertaste, and demographic inputs. The dataset is ideal for building classification models to understand sensory perception and explore the relationships between different flavor properties.

📋 Example Content: Features: taste intensity ratings, flavor profiles, participant info

Target: preference class or flavor rating label

Use Cases: sensory science, product development, personalized taste modeling

🧾 Column Descriptors | Column Name | Description | | -------------------- | ---------------------------------------------------------------------- | | Sweetness | Intensity rating of sweetness (e.g., on a 1–10 scale) | | Sourness | Perceived sourness level from participant input | | Bitterness | Participant rating of bitterness intensity | | Umami | Rating for savory or umami flavor presence | | Aroma | Intensity or pleasantness of the product's smell | | Texture | Mouthfeel or consistency (e.g., smooth, crunchy, thick) | | Aftertaste | Strength or quality of flavor that lingers after consumption | | Age | Age of participant (used for demographic analysis) | | Gender | Gender of the participant | | Region | Geographic or cultural background (optional for preference analysis) | | Overall_Preference | Target variable—overall rating or class label (e.g., Liked / Disliked) |

Datasets used to generate figures and sample runs in the SENTINEL application in the journal article "SENTINEL: A Shiny App for Processing and Analysis of Fenceline Sensor Data". This dataset is associated with the following publication: MacDonald, M., W. Champion, and E. Thoma. SENTINEL: A Shiny App for Processing and Analysis of Fenceline Sensor Data. ENVIRONMENTAL MODELLING & SOFTWARE. Elsevier Science, New York, NY, 189: 0, (2025).

Water quality depends on many factors. Among them some factors have direct effect in maintaining minimum sustainable environment. The following dataset contains parameters of three basic water quality factors. Temperature, pH factor and Water Turbidity. Two sets of Arduino based digital sensors were used to measure data from two different depths of the water label. Data were collected from a fish pond within the premises of Khan Jahan Ali Hall, Khulna University. Data were recorded 24 hours continuously from 15 January 2020 to 22 January 2020. Recording rate was 1 set of data per minute in average. All the sensors collected data in parallel at the same time. An Arduino Mega microcontroller board acted as the central processing unit and recorded all the sensors’ data altogether. Data file contents: Data 1:: Sensor data for 30 cm.xlsx: Contains raw data from pH sensor, Temperature sensor and Turbidity sensor. This set of sensors was 30 cm underwater. This file has 9623 sets of data each set having 3 data samples from the corresponding sensors. Data arrangement is described below: Column 1: Date and time of data recording in the format of YYYY-MM-DD [hh]:[mm]:[ss]. Time is in 24-hour format. Data are identical for all files. Column 2: Data from Temperature sensor in °C. Temperature values have maximum two decimal places. Column 3: Data from pH sensor. pH values have maximum two decimal places. Column 4: Data from turbidity sensor in “NTU” unit. Data 2:: Sensor data for 60 cm.xlsx: Contains raw data from Temperature sensor and Turbidity sensor. This set of sensors was 60 cm underwater. This file has 9623 sets of data each set having 2 data samples from the corresponding sensors. Data arrangement is described below: Column 1: Date and time of data recording, same as the previous file. Column 2: Data from Temperature sensor in °C. Temperature values have maximum two decimal places. Column 3: Data from turbidity sensor in “NTU” unit. Row-wise data categorisation: Each row represents one set of data for that consecutive time of the day. Considering when the data were taken the rows can be divided into two groups, day time data and night time data. Row 1-117, 811-1446, 2170-2856, 3634-4411, 5156-5887, 6554-7190, 7833-8477, 9078-9623:: Contains day time data from 6:00am to 6:00pm. Row 118-810, 1447-2169, 2857-3633, 4412-5155, 5888-6553, 7191-7832, 8478-9077:: Contains night time data from 6:00pm to 6:00am. These data can also be categorised row-wise based on the weather condition of the day. There are two groups of data that were collected under two different weather conditions. Row 1-6147, 7461-9623:: Contains sensor data from dry day. Row 6148-7460:: Contains sensor data from rainy day.

This dataset can be used to determine environmental suitability of the local area for fish farming. Machine learning based projects can use this data for detecting anomalies and forecasting near future aquatic environmental condition.

Purpose of the Notebook

This notebook aims to predict the HVAC system's power consumption (active_power) at a given time using the previous 15 minutes of sensor and operational data. For example, to predict the power at 10:00, the model uses data from 9:45 to 10:00. The notebook provides data cleaning, feature engineering, and modeling steps for this predictive task. Additionally, it may require further feature engineering and data wrangling to enhance model performance and data usability.

Dataset Description

This dataset contains 3 months of historical data from an HVAC system, with records every 5 minutes. The data includes operational parameters and environmental sensor readings, both inside and outside the cooled space.

Variables Description

• timestamp: Timestamp in the format YYYY-MM-DD HH:MM
• on_off: Equipment on/off sensor (0=off, 1=on)
• damper: Damper opening percentage (%)
• active_energy: Active energy (kWh, estimated)
• co2_1: CO2 sensor (ppm, estimated)
• ambient_humidity: Indoor relative humidity (%)
• active_power: Power consumption (Watts)
• power_generated: Generated power, if available (Watts, estimated)
• high_pressure_1: High pressure sensor 1 (bar)
• high_pressure_2: High pressure sensor 2 (bar)
• high_pressure_3: High pressure sensor 3 (bar)
• low_pressure_1: Low pressure sensor 1 (bar)
• low_pressure_2: Low pressure sensor 2 (bar)
• low_pressure_3: Low pressure sensor 3 (bar)
• outside_temp: Outside temperature (°C)
• outlet_temp: Outlet air temperature (°C)
• inlet_temp: Inlet air temperature (°C)
• summer_SP_temp: Summer setpoint temperature (°C)
• winter_SP_temp: Winter setpoint temperature (°C)
• ambient_temp: Indoor ambient temperature (°C)

Categorization is an essential task for sensory perception. Individuals learn category labels using a variety of strategies to ensure that sensory signals, such as sounds or images, can be assigned to proper categories. Categories are often learned on the basis of extreme examples, and the boundary between categories can differ among individuals. The trajectories for learning also differ among individuals, as different individuals rely on different strategies, such as repeating or alternating choices. However, little is understood about the relationship between individual learning trajectories and learned categorization. To study this relationship, we trained mice to categorize auditory stimuli into two categories using a two-alternative forced choice task. Because the mice took several weeks to learn the task, we were able to quantify the time course of individual strategies and how they relate to how mice categorize stimuli around the categorization boundary. Different mice exhibited ..., This dataset contains raw behavioral data collected from 19 mice performing an auditory two-alternative forced choice categorization task, in which they turned a Lego wheel clockwise or counterclockwise to report whether a presented tone burst stimuli was perceived as being drawn from a low- or high-frequency distribution. It also contains learning trajectories (parameterized by three evolving weights: "Low Category Knowledge", "High Category Knowledge" and "Choice Bias") extracted from the behavioral data using PsyTrack (Roy et al, 2021). It also contains fitting results from:

Fitting psychometric curves to testing sessions using PyBADS (Singh et al, 2023) Fitting two reinforcement learning models to training data , , # Mouse 2AFC Categorization: Open Source Data for Collina et al, "Individual-specific strategies in category learning inform learned boundaries"

This data, necessary to reproduce figures, is available at https://doi.org/10.5061/dryad.73n5tb359

Description of the data and file structure

This data summarizes the behavior of mice performing a 2AFC task in which they were trained to respond to categorical stimuli using a wheel. The trajectories of their learning of the associations between wheel turns and stimulus categories were extracted and studied, and a computational model was used to understand the factors that might explain individual variability in learning trajectory.

The data is divided into 5 folders:

1. MouseData

a. At the first level, MouseData is divided into folders based on the mouse ID. For example, the folder \"GS027\" contains all of the files pertaining to the behavior performed by the mouse GS027. Within that f...,

The data set presents the results of the sensory analysis of bread samples with the addition of beetroot, which was carried out by the sensory profiling analysis method based on the PN-ISO 11035: 1999 standard – “Sensory analysis - Identification and selection of descriptors for determining the sensory profile using multivariate methods”. The method was used to evaluate the maximum beet additive acceptable for a potential consumer. The description of the data within the file is in polish.

Variables extracted from studies used in systematic review