
🔬 Research & Academia · Flowchart
The experiment pipeline figure for a machine-learning paper: data, preprocessing, splits, model training, evaluation, ablation and statistical testing.
Drawing diagram…
Machine learning experiment pipeline for a paper: raw dataset is cleaned and preprocessed, split into train, validation and test sets, features are engineered, models are trained with a hyper-parameter search using the validation set, the best model is evaluated once on the test set, ablation studies check each component, and results are compared with baselines using statistical significance tests over several random seeds.
flowchart LR
D[("Raw dataset")] --> P["Clean and preprocess"]
P --> S{"Split"}
S --> TR["Train set"]
S --> VA["Validation set"]
S --> TE["Test set"]
TR --> FE["Feature engineering"]
FE --> MT["Train models"]
VA --> HS["Hyper-parameter search"]
MT --> HS
HS --> BM["Best model"]
BM --> EV["Evaluate once on test set"]
TE --> EV
EV --> AB["Ablation studies"]
EV --> BL["Compare with baselines<br/>several random seeds"]
BL --> ST["Significance tests"]
AB --> RS["Results and tables"]
ST --> RSThe standard flow diagram for a systematic review: records identified, screened, assessed for eligibility and included. Replace each n with your own counts.
The CONSORT flow diagram for a randomised trial: enrolment, allocation to two arms, follow-up and analysis, with the numbers lost at each stage.
Study-design figure for a randomised controlled trial: population, randomisation, two arms, measurements and outcomes.
Design figure for a prospective cohort study: exposed and unexposed groups followed over time to compare outcomes.
A conceptual framework with independent, mediating, moderating and dependent variables, plus control variables. Rename the boxes for your own study.
The six steps of thematic analysis (Braun and Clarke) as a process figure, from familiarisation to the written report.