Running Adjustments

Learn how to process your data and run turbulence intensity adjustments using TACT.

Quick Start

from tact import TACT
from tact.utils.load_data import load_data
from tact.utils.setup_processors import setup_processors

# 1. Load your data
data = load_data("your_data.csv")

# 2. Set up processors
binning_proc, ti_proc, stats_proc = setup_processors("config.json")

# 3. Process data
data = binning_proc.process(data)
data = ti_proc.process(data)

# 4. Run adjustment
tact = TACT()
results = tact.adjust(
    data=data,
    method="ss-sf",  # Recommended method
    parameters={"split": True, "config_path": "config.json"}
)

# 5. Access results
adjusted_data = results["adjusted_data"]
print(f"Adjusted {len(adjusted_data)} observations")

Step-by-Step Workflow

Load Data

Import your CSV file using TACT’s data loader:

from tact.utils.load_data import load_data

data = load_data("path/to/your/data.csv")
print(f"Loaded {len(data)} rows")
print(f"Columns: {data.columns.tolist()}")

What it does:

Reads CSV into pandas DataFrame
Validates column existence
Handles timestamps automatically

Set Up Processors

Create processors using your config file:

from tact.utils.setup_processors import setup_processors

binning_proc, ti_proc, stats_proc = setup_processors("config.json")

Returns three processors:

binning_proc: Bins data by wind speed
ti_proc: Calculates turbulence intensity metrics
stats_proc: Computes statistical summaries

Process Data

Apply processors to prepare data:

# Apply binning
data = binning_proc.process(data)

# Calculate TI metrics
data = ti_proc.process(data)

# Optional: Compute statistics
data = stats_proc.process(data)

Processing order matters! Always apply in this sequence:

Binning first
TI calculation second
Statistics last (optional)

Run Adjustment

Choose a method and run the adjustment:

from tact import TACT

tact = TACT()
results = tact.adjust(
    data=data,
    method="ss-sf",  # or "ssws", "sswsstd", "baseline"
    parameters={
        "split": True,              # Use train/test split
        "config_path": "config.json"
    }
)

Method options:

ss-sf: Recommended for most cases
sswsstd: Dual WS+SD adjustment
ssws: Wind speed only
baseline: No adjustment (reference)

Access Results

Extract the adjusted data and statistics:

# Get adjusted dataset
adjusted_data = results["adjusted_data"]

# View regression results (if applicable)
if "reg_results" in results:
    print(results["reg_results"])

# Check statistics
if "all_stats" in results:
    print(results["all_stats"])

Method Selection

SS-SF (Recommended)
SSWSStd
SSWS
Baseline

Site-Specific Simple + Filter

results = tact.adjust(
    data=data,
    method="ss-sf",
    parameters={"split": True, "config_path": "config.json"}
)

Best for:

General purpose turbulence adjustment
When you want the best MRBE/RRMSE
Avoiding error propagation

Performance: MRBE 37.8%, RRMSE 111.1%

Wind Speed + Standard Deviation

results = tact.adjust(
    data=data,
    method="sswsstd",
    parameters={"split": True, "config_path": "config.json"}
)

Best for:

Complex bias in both WS and SD
When you need independent WS and SD corrections
Large datasets (2000+ points)

Performance: MRBE 37.7%, RRMSE 121.6%

Wind Speed Only

results = tact.adjust(
    data=data,
    method="ssws",
    parameters={"split": True, "config_path": "config.json"}
)

Best for:

Simple wind speed bias correction
When SD measurements are less reliable
Simpler interpretation

Performance: MRBE 92.1%, RRMSE 188.0%

No Adjustment

results = tact.adjust(
    data=data,
    method="baseline",
    parameters={"config_path": "config.json"}
)

Best for:

Benchmarking other methods
Understanding raw measurement differences
Validation purposes

Performance: MRBE 90.4%, RRMSE 183.0%

Parameters Guide

split (Train/Test Split)

parameters={"split": True, "config_path": "config.json"}

What it does:

Splits data into training (70%) and testing (30%) sets
Trains regression on training data
Applies adjustment to test data

When to use:

For SS-SF, SSWS, SSWSStd methods
When you have sufficient data (500+ points)
❌ Not needed for baseline method

When split=False:

Uses all data for both training and testing
May lead to overfitting
Only recommended for small datasets

config_path (Configuration File)

parameters={"config_path": "path/to/config.json"}

Required for: All methodsWhat it contains:

Column name mappings
Binning configuration
Method-specific parameters

Path options:

Relative: "config.json" (same directory as script)
Absolute: "/full/path/to/config.json"

Method-Specific Parameters

Some methods accept additional parameters:

# Example: Custom parameters (if supported)
parameters={
    "split": True,
    "config_path": "config.json",
    "custom_param": value  # Method-specific
}

Check each method’s API documentation for available parameters.

Output Structure

The adjust() method returns a dictionary with these keys:

{
    "adjusted_data": pd.DataFrame,  # Main output - adjusted dataset
    "reg_results": pd.DataFrame,    # Regression statistics (if applicable)
    "all_stats": pd.DataFrame       # Summary statistics (if applicable)
}

Adjusted Data Columns

The returned DataFrame includes original columns plus:

Column	Description
`adjTI_RSD_TI`	Adjusted turbulence intensity
`adjRepTI_RSD_RepTI`	Adjusted representative TI
`RSD_adjWS`	Adjusted wind speed (SSWSStd only)
`RSD_adjSD`	Adjusted std deviation (SSWSStd only)

Save Results

# Save adjusted data
adjusted_data = results["adjusted_data"]
adjusted_data.to_csv("adjusted_results.csv", index=False)
print("Saved to adjusted_results.csv")

Common Workflows

Single Method
Compare Methods
Batch Processing

# Run one method and save results
from tact import TACT
from tact.utils.load_data import load_data
from tact.utils.setup_processors import setup_processors

# Load and process
data = load_data("data.csv")
bp, tp, sp = setup_processors("config.json")
data = bp.process(tp.process(data))

# Run SS-SF
tact = TACT()
results = tact.adjust(data, "ss-sf", {"split": True, "config_path": "config.json"})

# Save
results["adjusted_data"].to_csv("ss_sf_results.csv", index=False)

# Run and compare multiple methods
methods = ["baseline", "ss-sf", "ssws", "sswsstd"]
tact = TACT()

all_results = {}
for method in methods:
    results = tact.adjust(
        data=data,
        method=method,
        parameters={"split": True, "config_path": "config.json"}
    )
    all_results[method] = results

# Save each method's results
for method, results in all_results.items():
    results["adjusted_data"].to_csv(f"{method}_results.csv", index=False)

# Process multiple datasets
import glob

tact = TACT()
data_files = glob.glob("data/*.csv")

for file_path in data_files:
    # Load and process
    data = load_data(file_path)
    bp, tp, sp = setup_processors("config.json")
    data = bp.process(tp.process(data))

    # Run adjustment
    results = tact.adjust(data, "ss-sf", {"split": True, "config_path": "config.json"})

    # Save with descriptive name
    output_name = file_path.replace("data/", "results/").replace(".csv", "_adjusted.csv")
    results["adjusted_data"].to_csv(output_name, index=False)
    print(f"Processed {file_path}")

Next Steps

Validate Results

Run DNV RP-0661 validation

Visualize Data

Create plots and charts

Troubleshooting

Fix common issues

Getting Started

Guides

Advanced

Running Adjustments

Quick Start

Step-by-Step Workflow

Method Selection

Parameters Guide

Output Structure

Adjusted Data Columns

Save Results

Common Workflows

Next Steps

Validate Results

Visualize Data

Troubleshooting

Getting Started

Guides

Advanced

​Quick Start

​Step-by-Step Workflow

​Method Selection

​Parameters Guide

​Output Structure

​Adjusted Data Columns

​Save Results

​Common Workflows

​Next Steps

Validate Results

Visualize Data

Troubleshooting

Quick Start

Step-by-Step Workflow

Method Selection

Parameters Guide

Output Structure

Adjusted Data Columns

Save Results

Common Workflows

Next Steps