This dataset simulates anonymized mobile screen time and app usage data collected from Android/iOS users over a 3-month period (Jan–April 2024). It captures daily usage trends across various app categories including:

Productivity: Google Docs, Notion, Slack

Entertainment: YouTube, Netflix, TikTok

Social Media: Instagram, WhatsApp, Facebook

Utilities: Chrome, Gmail, Maps

For YouTube, additional engagement statistics such as views, likes, and comments are included to analyze video popularity and content consumption behavior.

The dataset enables exploration of:

Productivity vs. entertainment screen time patterns

Daily usage fluctuations

App-specific user engagement

Correlation between time spent and user interactions

YouTube content virality metrics

This is a great resource for:

EDA projects

Behavioral clustering

Dashboard development

Time series and anomaly detection

Building recommendation or focus-assistive apps

🚀 MobileViews: A Large-Scale Mobile GUI Dataset

MobileViews is a large-scale dataset designed to support research on mobile agents and mobile user interface (UI) analysis. The first release, MobileViews-600K, includes over 600,000 mobile UI screenshot-view hierarchy (VH) pairs collected from over 20,000 apps on the Google Play Store. This dataset is based on the DroidBot, which we have optimized for large-scale data collection, capturing more comprehensive interaction details… See the full description on the dataset page: https://huggingface.co/datasets/mllmTeam/MobileViews.

User Feedback Dataset from the Top 15 Downloaded Mobile Applications

This dataset comprises user feedback data collected from 15 globally acclaimed mobile applications, spanning diverse categories. The included applications are among the most downloaded worldwide, providing a rich and varied source for analysis. The dataset is particularly suitable for Natural Language Processing (NLP) applications, such as text classification and topic modeling.

List of Included Applications:

• TikTok
• Instagram
• Facebook
• WhatsApp
• Telegram
• Zoom
• Snapchat
• Facebook Messenger
• Capcut
• Spotify
• YouTube
• HBO Max
• Cash App
• Subway Surfers
• Roblox

Data Columns and Descriptions:

• review_id: Unique identifiers for each user feedback/application review.
• content: User-generated feedback/review in text format.
• score: Rating or star given by the user.
• TU_count: Number of likes/thumbs up (TU) received for the review.
• app_id: Unique identifier for each application.
• app_name: Name of the application.
• RC_ver: Version of the app when the review was created (RC).

Terms of Use:

This dataset is open access for scientific research and non-commercial purposes. Users are required to acknowledge the authors' work and, in the case of scientific publication, cite the most appropriate reference:

1.Paper

M. H. Asnawi, A. A. Pravitasari, T. Herawan, and T. Hendrawati, "The Combination of Contextualized Topic Model and MPNet for User Feedback Topic Modeling," in IEEE Access, vol. 11, pp. 130272-130286, 2023, doi: https://doi.org/10.1109/ACCESS.2023.3332644

2.Dataset

Asnawi, M. H., Pravitasari, A. A., Herawan, T., & hendrawati, T. (2023). User Feedback Dataset from the Top 15 Downloaded Mobile Applications [Data set]. In The Combination of Contextualized Topic Model and MPNet for User Feedback Topic Modeling (1.0.0, Vol. 11, pp. 130272–130286). Zenodo. https://doi.org/10.5281/zenodo.10204232

Researchers and analysts are encouraged to explore this dataset for insights into user sentiments, preferences, and trends across these top mobile applications. If you have any questions or need further information, feel free to contact the dataset authors.

Context

Data-driven models help mobile app designers understand best practices and trends, and can be used to make predictions about design performance and support the creation of adaptive UIs. This paper presents Rico, the largest repository of mobile app designs to date, created to support five classes of data-driven applications: design search, UI layout generation, UI code generation, user interaction modeling, and user perception prediction. To create Rico, we built a system that combines crowdsourcing and automation to scalably mine design and interaction data from Android apps at runtime. The Rico dataset contains design data from more than 9.3k Android apps spanning 27 categories. It exposes visual, textual, structural, and interactive design properties of more than 66k unique UI screens. To demonstrate the kinds of applications that Rico enables, we present results from training an autoencoder for UI layout similarity, which supports query-by-example search over UIs.

Content

Rico was built by mining Android apps at runtime via human-powered and programmatic exploration. Like its predecessor ERICA, Rico’s app mining infrastructure requires no access to — or modification of — an app’s source code. Apps are downloaded from the Google Play Store and served to crowd workers through a web interface. When crowd workers use an app, the system records a user interaction trace that captures the UIs visited and the interactions performed on them. Then, an automated agent replays the trace to warm up a new copy of the app and continues the exploration programmatically, leveraging a content-agnostic similarity heuristic to efficiently discover new UI states. By combining crowdsourcing and automation, Rico can achieve higher coverage over an app’s UI states than either crawling strategy alone. In total, 13 workers recruited on UpWork spent 2,450 hours using apps on the platform over five months, producing 10,811 user interaction traces. After collecting a user trace for an app, we ran the automated crawler on the app for one hour.

Acknowledgements

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN https://interactionmining.org/rico

Inspiration

The Rico dataset is large enough to support deep learning applications. We trained an autoencoder to learn an embedding for UI layouts, and used it to annotate each UI with a 64-dimensional vector representation encoding visual layout. This vector representation can be used to compute structurally — and often semantically — similar UIs, supporting example-based search over the dataset. To create training inputs for the autoencoder that embed layout information, we constructed a new image for each UI capturing the bounding box regions of all leaf elements in its view hierarchy, differentiating between text and non-text elements. Rico’s view hierarchies obviate the need for noisy image processing or OCR techniques to create these inputs.

We surveyed 10,208 people from more than 15 countries on their mobile app usage behavior. The countries include USA, China, Japan, Germany, France, Brazil, UK, Italy, Russia, India, Canada, Spain, Australia, Mexico, and South Korea. We asked respondents about: (1) their mobile app user behavior in terms of mobile app usage, including the app stores they use, what triggers them to look for apps, why they download apps, why they abandon apps, and the types of apps they download. (2) their demographics including gender, age, marital status, nationality, country of residence, first language, ethnicity, education level, occupation, and household income (3) their personality using the Big-Five personality traits This dataset contains the results of the survey.

Mobile app usage events with app names and categories, available globally and joinable by device ID.

📱 Mobile App Usage & Screen Time (2022–2024)

This synthetic dataset captures global mobile app usage behavior and screen time patterns across 2022 to 2024. It contains 10,000 usage records across 15 countries covering 80+ popular apps across 12 categories including Social Media, Gaming, Streaming, Productivity, Health & Fitness, Education, Shopping, Finance, and more.

📊 Dataset Highlights

• 10,000 records × 34 features — zero null values
• 80+ popular apps across 12 categories
• 15 countries | Android & iOS | 10 device brands
• Covers usage, monetization, notifications & digital wellbeing

🎯 Key Research Areas

• ⏱️ Screen time analysis by age group, gender & country
• 📱 App category engagement and session duration patterns
• 🔔 Notification overload impact on usage behavior
• 💰 In-app purchase prediction using ML classification
• 😴 Sleep disruption correlation with late-night phone use
• 🧠 Mental health impact of social media & gaming
• 🌿 Digital wellbeing feature adoption rates
• 📈 Screen time growth trend from 2022 to 2024

📋 Feature Categories

⚠️ Disclaimer

Fully synthetic dataset. No real user data was collected. For educational and research purposes only.

Author: Hamna Munir | License: CC0 Public Domain | Year: 2026

This is a dataset of the paper titled "Strategies to Embed Human Values in Mobile Apps: What do End-Users and Practitioners Think?". In this study, we conducted a mixed-methods empirical study, which collected data through 13 semi-structured interviews with Bangladeshi agriculture mobile app practitioners and 4 focus groups with 20 Bangladeshi female farmers. Our aim is to identify the extent to which the existing agriculture mobile apps reflect Bangladeshi female farmers' values and to propose potential strategies to address their values in agriculture apps. There are five documents in this dataset. Questionnaire for focus groups Questionnaire for interviews Examples of open coding process Summary of member checking outcomes Illustrative quotations for values and strategies

This repository is part of the ITC-NetMingledApp dataset, which includes network traffic data from 36 Android applications, with each capture featuring concurrent traffic from multiple applications and smartphones. This repository contains part #1 of the data related to the Iran-Tehran scenario. Each capture is stored in a compressed file containing the relevant PCAP files of the associated applications. The PCAP files are named according to a convention: {TimeStamp}_{Application Name}{Download-Upload Speed}.pcap Part #2 of Iran-Tehran scenario is in the Tehran Dataset #2 (https://doi.org/10.17632/zsffy3j9y6.1) repository.

Complete dataset included in the full report. Detailed tables, regional splits, forecasts, and methodologies are available with purchase.

These data have been collected and supplied by Huq Ltd. and comprise of records for the period July 2016 to October 2020. The data contain aggregated geolocated activity counts derived from mobile phone app use across Great Britain.

Mobile phone applications seek user’s consent for recording and storing the mobile device’s location when the app is in use. Activity counts are derived from these locations as the sum of distinct devices per grid cell per day. These data can be used as proxy for estimating activity levels and footfall across the UK.

These aggregate data were created from record level data which comprised individual phone IDs, and multiple entries for each mobile device if it is used multiple times for one app or the user accesses multiple apps. Thus, the following data cleaning and aggregation process has been used to create the activity counts:

1. Cleaning: Daily records comprise unique device ID, time-stamp and location of each entry collected by any app. The time-stamp is reformatted as a single daily date attribute.

2. Spatial linkage to OSGB grid: After turning the daily data-frames into spatial objects, the files are joined to the 1km x 1km OSGB grid, and each impression is attributed a grid cell ID corresponding to its latitude and longitude.

3. Creation of activity counts: Activity counts are created following the previous steps by counting the number of unique device IDs per grid cell per date. This removes multiple appearances of the same device (one device may collect multiple impressions through different apps or due to frequent usage). The final activity count corresponds to the number of unique devices within a 1km square for that day.

4. Output: The output comprises cleaned aggregation counts for each grid cell and day

N.B. More detail on how the data was collected and coverage is available if requesting for this detail in your initial application purpose, or if contacting us by email once you have made your initial application and received the form. Applicants would need to sign a non-disclosure agreement before accessing this detail, and such as request will significantly increase the time for data delivery. You can, of course, make a full application for the data without first receiving this collection/ coverage metadata.

Content

These data are provided at 1km x 1km OSGB Grid cells.

Activity counts of 1-10 devices are masked and replaced by “*” in the database, as low counts present potentially identifiable information.

For detailed description of the columns contained within the data, see the Variable Dictionary; and for an overview of the characteristics of the data, see the Data Summary. These files can be downloaded from the bottom of this page.

Quality, Representation and Bias

Excellent quality and coverage for major towns and cities. The data may be less complete for smaller settlements or more rural areas. Data are subject to suppression of potentially disclosive low counts as detailed above. Huq collects data from a varying mix of apps, the identities of which are commercially sensitive. Apps may be added to or deleted from the secure and summary data products over time. This, along with increasing national coverage and mobile phone uptake, results in general increases in apparent activity over the period covered by the data.

The dataset would benefit from comparison with population estimates (e.g. census data) to investigate coverage issues. 2016 data have the highest percentage of suppressed counts, and data suppression generally decreases over time, particularly in metropolitan (Met) areas. Data suppression levels in metropolitan areas generally fall below 50% by 2020.

Complete dataset included in the full report. Detailed tables, regional splits, forecasts, and methodologies are available with purchase.

Mobile app statistics: 2,172,472 App Store apps, 929,754,061 weekly downloads and a $1.4 trillion app economy, from primary sources.

MobileWorld: Benchmarking Autonomous Mobile Agents in Agent-User Interactive and MCP-Augmented Environments

Mobile World is a substantially more challenging mobile-use benchmark designed to better reflect real-world mobile usage. It comprises 201 tasks across 20 applications, featuring long-horizon, cross-app tasks, and novel task categories including agent-user interaction and MCP-augmented tasks. The difficulty of Mobile World is twofold:

Long-horizon, cross-application… See the full description on the dataset page: https://huggingface.co/datasets/Tongyi-MAI/MobileWorld.

App store data from Apple App Store and Google Play — rankings, ratings, reviews, categories and metadata. Mobile app and ASO intelligence, ready or custom.

This dataset provides a detailed sentiment analysis of user reviews for the app, alongside comprehensive app information from the Google Play Store. The first dataset (App_Sentiment_Analysis.csv) includes translated review texts, sentiment classifications, and numerical scores for sentiment polarity and subjectivity, offering insights into user opinions and experiences. The second dataset (Review.csv) covers various attributes of several apps, such as their ratings, review counts, sizes, installation numbers, content ratings, genres, and more. Together, these datasets facilitate a thorough analysis of user feedback and app performance, supporting app recommendation and improvement strategies.

This is a dataset of the paper titled "Strategies to Embed Human Values in Mobile Apps: What do End-Users and Practitioners Think?". In this study, we conducted a mixed-methods empirical study, which collected data through 13 semi-structured interviews with Bangladeshi agriculture mobile app practitioners and 4 focus groups with 20 Bangladeshi female farmers. Our aim is to identify the extent to which the existing agriculture mobile apps reflect Bangladeshi female farmers' values and to propose potential strategies to address their values in agriculture apps. There are four documents in this dataset. Questionnaire for focus groups Questionnaire for interviews Examples of open coding process Summary of member checking outcomes Illustrative quotations for values and strategies

This dataset provides detailed, event-level records of mobile app feature usage, including user interactions, device context, session information, and user segmentation. It enables product teams and UX researchers to analyze feature adoption rates, engagement patterns, and user cohorts, supporting data-driven decisions for app improvement and user experience optimization.

PARROT2025_mitmproxy Dataset: Mobile App Traffic Traces with SSL Keys

Dataset Overview:

• Traffic from 80 Android applications selected from the MAppGraph dataset app list
• 4 traffic traces per application
• Average duration of 5 minutes per trace
• Total of 320 traffic captures spanning more than 25 hours of app traffic

Traffic Characteristics:

Each trace captures the network traffic generated during the first deployment of the application following installation. The traffic represents app initialization behavior and startup communications patterns.

Capture System Configuration:

• Traffic captured using the PARROT capture system with mitmproxy traffic interception
• Android SDK API level 35 (Android 15) emulator
• DNS configuration: dns.google (8.8.8.8)
• Capture location: Universidad Carlos III de Madrid, Spain
• Capture year: 2025
• Traffic generation: Manual interaction with the emulator by research team members

Data Format and Organization:

The dataset is contained in the folder "dataset_PARROT2025_mitmproxy" which includes:

• Raw network traffic traces in PCAP format
• Corresponding SSL session keys for traffic decryption

Research Applications:

This dataset enables research in mobile app traffic analysis, network security, protocol evolution studies, and encrypted traffic analysis. The inclusion of SSL keys allows researchers to analyze both encrypted and decrypted communications patterns.

Design teams create and refine mobile app prototypes, allowing users to visualize their ideas before development. This iterative process involves managing various assets such as fonts, device templates, and component instances, while fostering collaboration among workspace members. Each project presents a unique blend of complexity, with some prototypes containing numerous layers and features, while others maintain a more straightforward approach. User engagement metrics and the frequency of shared prototypes become vital indicators of success and team collaboration.

In this dataset, the tables provide a structured overview of the operations within the design team. The 'projects' table represents individual app prototypes, detailing their characteristics and statuses. The 'teams' table captures the various groups working on these projects, highlighting the members involved and their roles. Resources necessary for the development of prototypes, such as fonts and templates, are documented in the 'resources' table, linking to specific projects and teams. The 'user_engagement' table reveals how users interact with the prototypes, offering insights into usage patterns and engagement levels. Lastly, the 'prototype_layers' table details the components of each prototype, breaking down the various features and complexities involved.

Joining these tables allows for comprehensive analysis of how team dynamics influence project outcomes and user engagement. For instance, by connecting the 'projects' with 'user_engagement', developers can explore how different prototypes attract user interactions based on their design intricacies. Similarly, analyzing 'teams' alongside 'projects' can provide insights into the effectiveness of group collaboration on prototype success. This dataset answers pertinent questions about project performance, team efficiency, and resource utilization, making it an excellent resource for developers looking to practice data analysis in a SaaS environment. It is important to note that the data is synthetic, crafted to simulate real-world scenarios in mobile app prototyping.