All of the datasets and the below description are quoted from Project - Data-driven prediction of battery cycle life before capacity degradation.

Objective

This dataset, used in our publication “Data-driven prediction of battery cycle life before capacity degradation”, consists of 124 commercial lithium-ion batteries cycled to failure under fast-charging conditions. These lithium-ion phosphate (LFP)/graphite cells, manufactured by A123 Systems (APR18650M1A), were cycled in horizontal cylindrical fixtures on a 48-channel Arbin LBT potentiostat in a forced convection temperature chamber set to 30°C. The cells have a nominal capacity of 1.1 Ah and a nominal voltage of 3.3 V.

The objective of this work is to optimize fast charging for lithium-ion batteries. As such, all cells in this dataset are charged with a one-step or two-step fast-charging policy. This policy has the format “C1(Q1)-C2”, in which C1 and C2 are the first and second constant-current steps, respectively, and Q1 is the state-of-charge (SOC, %) at which the currents switch. The second current step ends at 80% SOC, after which the cells charge at 1C CC-CV. The upper and lower cutoff potentials are 3.6 V and 2.0 V, respectively, which are consistent with the manufacturer’s specifications. These cutoff potentials are fixed for all current steps, including fast charging; after some cycling, the cells may hit the upper cutoff potential during fast charging, leading to significant constant-voltage charging. All cells discharge at 4C.

The dataset is divided into three “batches”, representing approximately 48 cells each. Each batch is defined by a “batch date”, or the date the tests were started. Each batch has a few irregularities, as detailed on the page for each batch.

The data is provided in two formats. For each batch, a MATLAB struct is available. The struct provides a convenient data container in which the data for each cycle is easily accessible. This struct can be loaded in either MATLAB or python (via the h5py package). Pandas dataframes can be generated via the provided code. Additionally, the raw data for each cell is available as a CSV file. Note that the CSV files occasionally exhibit errors in both test time and step time in which the test time resets to zero mid-cycle; these errors are corrected for in the structs.

The temperature measurements are performed by attaching a Type T thermocouple with thermal epoxy (OMEGATHERM 201) and Kapton tape to the exposed cell can after stripping a small section of the plastic insulation. Note that the temperature measurements are not perfectly reliable; the thermal contact between the thermocouple and the cell can may vary substantially, and the thermocouple sometimes loses contact during cycling.

Internal resistance measurements were obtained during charging at 80% SOC by averaging 10 pulses of ±3.6C with a pulse width of 30 ms (2017-05-12 and 2017-06-30) or 33 ms (2018-04-12).

The following repository contains some starter code to load the datasets in either MATLAB or python:

https://github.com/rdbraatz/data-driven-prediction-of-battery-cycle-life-before-capacity-degradation

Low rate data used to generate figure 4:

• 2018-02-20_batchdata_updated_struct_errorcorrect.mat
• 2018-04-03_varcharge_batchdata_updated_struct_errorcorrect.mat

If using this dataset in a publication, please cite: Severson et al. Data-driven prediction of battery cycle life before capacity degradation. Nature Energy volume 4, pages 383–391 (2019).

**Batch - 2017-05-12**
Experimental design
- All cells were cycled with one-step or two-step charging policies. The charging time varies from ~8 to 13.3 minutes (0-80% SOC). There are generally two cells tested per policy, with the exception of 3.6C(80%).
- 1 minute and 1 second rests were placed after reaching 80% SOC during charging and after discharging, respectively.
-We cycle to 80% of nominal capacity (0.88 Ah).
- An initial C/10 cycle was performed in the beginning of each test.
- The cutoff currents for the constant-voltage steps were C/50 for both charge and discharge.
- The pulse width of the IR test is 30 ms.

Experimental notes
- The computer automatically restarted twice. As such, there are some time gaps in the data.
- The temperature control is somewhat inconsistent, leading to variability in the baseline chamber temperature.
- The tests in channels 4 and 8 did not successfully start and thus do not have data.
- The thermocouples for channels 15 and 16 were accidentally switched.

Data notes
- Cycle 1 data is not available in the struct. The sampling rate for this cycle was initially too high, so we excluded it from the data set to create more manageable file sizes.
- The cells in Channels 1, 2, 3, 5, and 6 (3.6C(80%) and 4C(80%) policies) were stopped at the end of this batch and resume...

This dataset is the result of a project to support policy making by providing insights on the availability and composition of education offer in four key digital domains: artificial intelligence, high performance computing, cybersecurity, and data science. Following a text mining methodology that captures the inclusion of advanced digital technologies in the programmes’ syllabus, we monitor the availability of masters’ programmes, bachelor’s programmes and short professional courses and study their characteristics. These include the scope or depth with which the digital content is taught (classified into broad or specialised), education fields in which digital technologies are embedded (e.g., Information and communication technologies, Business, administration and law), and the content areas covered by the programmes (e.g. robotics, machine learning). Also, we consider the overlap between the four domains, to identify complementarities and synergies in the academic offer of advanced digital technologies. The dataset covers yearly data, starting from the academic year 2019-2020 and ending in academic year 2023-24 (and will not be further updated). In order to provide comparison with other competing economies, the dataset covers the EU and its Member States plus six additional countries: the United Kingdom, Norway, Switzerland, Canada, the United States, and Australia. Results of the study have been used as reference in the European Artificial Intelligence Strategy, the White Paper on Artificial Intelligence – a European approach to excellence and trust, in the Stanford University’s Artificial Intelligence Index Report 2019 and 2021. These data have substantiated the assessment of the national Recovery and Resilience plans, and are used as input for the Digital Resilience Dashboard, among others.

Abstract

The Longitudinal Aging Study in India (LASI) aims to understand the situation of India’s elderly population by collecting data on their health, social situations, and economic circumstances. It will provide a foundation for innovative, rigorous, and multidisciplinary studies of aging in India that will inform policy and advance scientific knowledge. Its goal is to provide data harmonized with the Health and Retirement Study (HRS) and its sister studies around the world. A pilot study has been conducted that includes household survey data, Computer-Assisted Personal Interviews (CAPI) and molecular biomarkers. The results of the pilot study will inform the design of a full-scale, nationally representative LASI, with a sample of roughly 30,000 to be followed longitudinally (with refresher populations added as needed). Due to its harmonized design with parallel international studies, LASI will contribute to scientific insights and policy development in other countries as well. LASI will ultimately be part of a worldwide effort aimed at understanding how different institutions, cultures, and policies can understand and prepare for population ageing.

Documentation

You can download the pilot data at the Harvard Program on the Global Demography of Aging website

Methodology

The LASI pilot survey targeted 1,600 individuals aged 45 and older and their spouses, and will inform the design and rollout of a full-scale, nationally representative LASI survey. The expectation is that LASI will be a biennial survey and will be representative of Indians aged 45 and older, with no upper age limit.

1,600 age-qualifying individuals were drawn from a stratified, multistage area probability sampling design. After a series of pre-pilot studies designed to test the instrument and the key ideas behind it, pilot data were collected through face-to-face interviews over three month time periods. Descriptive analyses of the data will be performed and lessons will be drawn to inform the launching of a full-scale LASI survey.

The LASI pilot survey was conducted in four states: Karnataka, Kerala, Punjab, and Rajasthan. To capture regional variation we have included two northern states (Punjab and Rajasthan) and two southern states (Karnataka and Kerala). Karnataka and Rajasthan were included in the Study on Global AGEing and Adult Health (SAGE), which will enable us to compare our findings with the SAGE data. The inclusion of Kerala and Punjab demonstrates our aim to obtain a broader representation of India, where geographic variations accompanied by socioeconomic and cultural differences call for careful study and deliberation. Punjab is an example of an economically developed state, while Rajasthan is relatively poor, with very low female literacy, high fertility, and persisting gender disparities. Kerala, which is known for its relatively efficient health care system, has undergone rapid social development and is included as a potential harbinger of how other Indian states might evolve.

Abstract

The Longitudinal Indian Family Health (LIFE) is a long-term research study that will examine socio-economic and environmental influences on children’s health and development in India. The main aim of the study is to understand the link between the environmental conditions in which Indian women conceive, go through their pregnancy and give birth, and their physical and mental health during this period.

The cohort comprises married women between 15 and 35 years of age (mean 22 years), recruited before pregnancy or in the first trimester of pregnancy, from 2009 to 2011. These CHVs focus on women in the village to ascertain pregnancy (by interview) and to educate and encourage the women to seek regular antenatal care and other health care services. REACH has enumerated all household members in these communities and mapped each dwelling by a geographical information system (GIS). During each visit, CHVs conduct interviews to collect and update information on demography and pregnancy. Since 2004, CHVs have been collecting data on infant deaths and birthweights in the population. Socio-demographic variables such as access to electricity, means of transportation and possession of audio-visual devices were collected from REACH database

Documentation

You can submit a proposal to collaborate with LIFE Study investigators. A written protocol must be submitted, reviewed and approved by the LIFE Data Sharing Plan Committee before initiation of new projects. For further information, contact Dr P. S. Reddy at [reddyps@verizon.net]. Updated information may be found on the research centre website at [www.sharefoundations.org].

Methodology

The LIFE study is being conducted in villages of Medchal Mandal, R.R.District, Telangana, India. Since 2009, 1227 women aged between 15 and 35 years were recruited before conception or within 14 weeks of gestation. Women were followed through pregnancy, delivery, and postpartum. Follow-up of children is ongoing. Baseline data were collected from husbands of 642 women.

Anthropometric measurements, biological samples and detailed questionnaire data were collected during registration, the first and third trimesters, delivery and at 1 month postpartum. Anthropometric measurements and health questionnaire data are obtained for each child, and a developmental assessment is done at 1, 6, 12, 18, 24, 36, 48 and 60 months. At 36 months, each child is screened for development and mental health problems. Questionnaires are completed for pregnancy loss and death of children under 5 years old. The LIFE Biobank preserves over 6000 samples.

A seven-year randomized evaluation suggests education subsidies reduce adolescent girls' dropout, pregnancy, and marriage but not sexually transmitted infection (STI). The government's HIV curriculum, which stresses abstinence until marriage, does not reduce pregnancy or STI. Both programs combined reduce STI more, but cut dropout and pregnancy less, than education subsidies alone. These results are inconsistent with a model of schooling and sexual behavior in which both pregnancy and STI are determined by one factor (unprotected sex), but consistent with a two-factor model in which choices between committed and casual relationships also affect these outcomes. This data was collected as a part of the study "Education, HIV, and Early Fertility: Experimental Evidence from Kenya." Details on sample construction and data collection for this survey data can be found in the paper. The 2012 version of the paper is available here: http://www.stanford.edu/~pdupas/DDK_EducFertHIV.pdf. Note that all sections of data collected for the study are not currently available and will be released in the future.

The research data of this study were collected from live-streaming sessions on global e-commerce platforms, including Amazon Live and YouTube Shopping. These broadcasts feature professional English-speaking hosts with established credibility in the e-commerce field, focusing primarily on apparel and home furnishings categories. For analytical purposes, over 400 minutes of live-streaming were captured, subsequently transcribed into textual format, and meticulously verified through manual review.This study built a dependency treebank specifically designed for English e-commerce live-streaming texts (punctuation excluded). A dependency treebank represents a syntactically annotated corpus based on dependency grammar. Within the framework of word grammar—a particular dependency grammar approach—the fundamental unit is the individual word, with sentence structure being entirely composed of categorized relationships between these words (Hudson, 2010). These asymmetric syntactic connections, known as dependency relations, link a governing word, the head, with its dependent. The treebank explicitly marks these heads, dependents, and their dependency relations.The syntactic dependency analysis of the corpora is conducted using Stanford Parser (Version 4.2.0), a tool developed by Stanford University’s Natural Language Processing research team. This software, available for download at https://nlp.stanford.edu/software/lex-parser.shtml#Download, has demonstrated superior annotation accuracy compared to similar tools.