Exoplanet Discovery Terminal

# All confirmed planets (PS table)
https://exoplanetarchive.ipac.caltech.edu/TAP/sync?query=select+*+from+ps+where+default_flag=1&format=json

# HZ candidates (insolation filter)
https://exoplanetarchive.ipac.caltech.edu/TAP/sync?query=select+pl_name,pl_rade,pl_insol+from+ps+where+pl_insol+between+0.25+and+1.5&format=csv

# TESS confirmed planets
https://exoplanetarchive.ipac.caltech.edu/TAP/sync?query=select+*+from+ps+where+disc_facility+like+'TESS%25'&format=json

import requests
import pandas as pd
from io import StringIO

# NASA Exoplanet Archive TAP endpoint
TAP_URL = "https://exoplanetarchive.ipac.caltech.edu/TAP/sync"

# ADQL query: Earth-sized planets in the habitable zone
query = """
SELECT pl_name, pl_rade, pl_insol, pl_orbper, st_teff, sy_dist,
       disc_year, discoverymethod, disc_facility
FROM ps
WHERE default_flag = 1
  AND pl_rade BETWEEN 0.5 AND 2.0
  AND pl_insol BETWEEN 0.25 AND 1.5
ORDER BY pl_insol DESC
"""

params = {
    "query": query,
    "format": "csv"
}

response = requests.get(TAP_URL, params=params, timeout=30)
response.raise_for_status()

df = pd.read_csv(StringIO(response.text))
print(f"Found {len(df)} habitable-zone rocky planets")
print(df[['pl_name', 'pl_rade', 'pl_insol', 'disc_year']].head(20))

import lightkurve as lk
import matplotlib.pyplot as plt

# Search for TESS light curves — TOI-700 (all sectors)
search_result = lk.search_lightcurve(
    "TOI-700",
    mission="TESS",
    author="SPOC"  # Science Processing Operations Center pipeline
)
print(search_result)

# Download all available sectors
lc_collection = search_result.download_all()

# Stitch sectors together and normalize
lc = lc_collection.stitch()
lc = lc.normalize().remove_outliers(sigma=5)

# Fold on TOI-700d period (37.42 days)
period_d = 37.4219  # days
t0 = 1325.3        # BTJD reference epoch

folded = lc.fold(period=period_d, epoch_time=t0)
folded.scatter(column='flux', s=0.1, alpha=0.5)
plt.title("TOI-700d Phase-Folded Transit Light Curve (TESS)")
plt.xlabel("Phase")
plt.ylabel("Normalized Flux")
plt.tight_layout()
plt.show()

import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split

# Load Kepler cumulative KOI table
df = pd.read_csv("cumulative_koi.csv")

# Target label: 1 = planet (CONFIRMED/CANDIDATE), 0 = false positive
df["label"] = (df["koi_disposition"] != "FALSE POSITIVE").astype(int)

# Select predictive features (from ExoMiner feature importance analysis)
FEATURES = [
    "koi_period", "koi_ror", "koi_srho", "koi_impact",
    "koi_duration", "koi_depth", "koi_teq", "koi_insol",
    "koi_steff", "koi_srad", "koi_slogg", "koi_model_snr",
    "koi_count"
]

# Drop rows with high null rate
threshold = 0.8  # keep rows with ≥80% feature coverage
df_clean = df.dropna(subset=FEATURES, thresh=int(len(FEATURES) * threshold))

X = df_clean[FEATURES].copy()
y = df_clean["label"].copy()

# Log-transform skewed features
log_cols = ["koi_period", "koi_depth", "koi_model_snr"]
X[log_cols] = np.log1p(X[log_cols])

# Impute remaining NaNs with median
X = X.fillna(X.median())

# Standardize
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# Stratified split (preserve class ratios — ~60% FP)
X_train, X_test, y_train, y_test = train_test_split(
    X_scaled, y, test_size=0.2, stratify=y, random_state=42
)

print(f"Training set: {X_train.shape[0]} samples")
print(f"Test set:     {X_test.shape[0]} samples")
print(f"Class balance — Train: {y_train.mean():.2%} positive")