一些小修改
This commit is contained in:
76
GPUMD/Umap/umap_make.py
Normal file
76
GPUMD/Umap/umap_make.py
Normal file
@@ -0,0 +1,76 @@
|
||||
from pathlib import Path
|
||||
import numpy as np
|
||||
import matplotlib
|
||||
matplotlib.use("Agg") # 仅保存图片，不弹窗
|
||||
import matplotlib.pyplot as plt
|
||||
from umap import UMAP
|
||||
|
||||
def umap_dir_to_pngs(dir_path: str) -> None:
|
||||
"""
|
||||
对目录内每个 .npy 文件执行 UMAP(30D->2D) 并保存散点图。
|
||||
- 输入 .npy 期望形状为 (n_samples, 30) 或 (30, n_samples)
|
||||
- 输出图片保存在同目录，命名为 <原文件名>_umap.png
|
||||
"""
|
||||
p = Path(dir_path)
|
||||
if not p.is_dir():
|
||||
raise ValueError(f"{dir_path!r} 不是有效文件夹")
|
||||
|
||||
files = sorted(p.glob("*.npy"))
|
||||
if not files:
|
||||
print(f"目录 {p} 中未找到 .npy 文件")
|
||||
return
|
||||
|
||||
for f in files:
|
||||
try:
|
||||
data = np.load(f)
|
||||
if data.ndim == 2:
|
||||
if data.shape[1] == 30:
|
||||
X = data
|
||||
elif data.shape[0] == 30:
|
||||
X = data.T
|
||||
else:
|
||||
print(f"[跳过] {f.name}: shape={data.shape}, 未检测到 30 维特征")
|
||||
continue
|
||||
else:
|
||||
print(f"[跳过] {f.name}: 期望二维数组，实际 shape={data.shape}")
|
||||
continue
|
||||
|
||||
# 清理非数值行
|
||||
mask = np.isfinite(X).all(axis=1)
|
||||
if not np.all(mask):
|
||||
X = X[mask]
|
||||
print(f"[提示] {f.name}: 移除了含 NaN/Inf 的样本行")
|
||||
|
||||
n_samples = X.shape[0]
|
||||
if n_samples < 3:
|
||||
print(f"[跳过] {f.name}: 样本数过少(n={n_samples})，无法稳定降维")
|
||||
continue
|
||||
|
||||
# 确保 n_neighbors 合法
|
||||
n_neighbors = min(15, max(2, n_samples - 1))
|
||||
reducer = UMAP(
|
||||
n_components=2,
|
||||
n_neighbors=n_neighbors,
|
||||
min_dist=0.1,
|
||||
metric="euclidean",
|
||||
random_state=42,
|
||||
)
|
||||
emb = reducer.fit_transform(X)
|
||||
|
||||
fig, ax = plt.subplots(figsize=(6, 5), dpi=150)
|
||||
ax.scatter(emb[:, 0], emb[:, 1], s=6, c="#1f77b4", alpha=0.8, edgecolors="none")
|
||||
ax.set_title(f"{f.name} • UMAP (n={len(X)}, nn={n_neighbors})", fontsize=10)
|
||||
ax.set_xlabel("UMAP-1")
|
||||
ax.set_ylabel("UMAP-2")
|
||||
ax.grid(True, linestyle="--", linewidth=0.3, alpha=0.5)
|
||||
fig.tight_layout()
|
||||
|
||||
out_png = f.with_suffix("").as_posix() + "_umap.png"
|
||||
fig.savefig(out_png)
|
||||
plt.close(fig)
|
||||
print(f"[完成] {f.name} -> {out_png}")
|
||||
except Exception as e:
|
||||
print(f"[错误] 处理 {f.name} 失败: {e}")
|
||||
|
||||
if __name__=="__main__":
|
||||
umap_dir_to_pngs("data")
|
||||
161
GPUMD/Umap/umap_make_2.py
Normal file
161
GPUMD/Umap/umap_make_2.py
Normal file
@@ -0,0 +1,161 @@
|
||||
from pathlib import Path
|
||||
import numpy as np
|
||||
import matplotlib
|
||||
matplotlib.use("Agg")
|
||||
import matplotlib.pyplot as plt
|
||||
|
||||
try:
|
||||
from umap import UMAP
|
||||
except Exception:
|
||||
from umap.umap_ import UMAP
|
||||
|
||||
|
||||
def umap_dir_shared_coords(
|
||||
dir_path: str,
|
||||
*,
|
||||
metric: str = "cosine",
|
||||
n_neighbors: int = 15,
|
||||
min_dist: float = 0.0,
|
||||
spread: float = 1.2,
|
||||
standardize: bool = False,
|
||||
context: bool = True,
|
||||
make_joint: bool = True,
|
||||
init: str = "random", # 关键：禁用谱初始化，避免告警；也可用 "pca"
|
||||
jitter: float = 0.0, # 可选：拟合前加微弱噪声，如 1e-6
|
||||
random_state: int = 42
|
||||
) -> None:
|
||||
"""
|
||||
在同一 UMAP 坐标系中为目录内每个 .npy 文件生成 2D 图。
|
||||
- 每个 .npy 形状为 (n_samples, 30) 或 (30, n_samples)
|
||||
- 统一坐标轴范围；各自输出 *_umap_shared.png，另可输出总览图
|
||||
"""
|
||||
p = Path(dir_path)
|
||||
if not p.is_dir():
|
||||
raise ValueError(f"{dir_path!r} 不是有效文件夹")
|
||||
|
||||
files = sorted(p.glob("*.npy"))
|
||||
if not files:
|
||||
print(f"目录 {p} 中未找到 .npy 文件")
|
||||
return
|
||||
|
||||
X_list, paths, counts = [], [], []
|
||||
for f in files:
|
||||
try:
|
||||
data = np.load(f)
|
||||
if data.ndim != 2:
|
||||
print(f"[跳过] {f.name}: 期望二维数组，实际 shape={data.shape}")
|
||||
continue
|
||||
|
||||
if data.shape[1] == 30:
|
||||
X = data
|
||||
elif data.shape[0] == 30:
|
||||
X = data.T
|
||||
else:
|
||||
print(f"[跳过] {f.name}: shape={data.shape}, 未检测到 30 维特征")
|
||||
continue
|
||||
|
||||
mask = np.isfinite(X).all(axis=1)
|
||||
if not np.all(mask):
|
||||
X = X[mask]
|
||||
print(f"[提示] {f.name}: 移除了含 NaN/Inf 的样本行")
|
||||
|
||||
if X.shape[0] < 3:
|
||||
print(f"[跳过] {f.name}: 样本数过少(n={X.shape[0]})")
|
||||
continue
|
||||
|
||||
X_list.append(X)
|
||||
paths.append(f)
|
||||
counts.append(X.shape[0])
|
||||
except Exception as e:
|
||||
print(f"[错误] 读取 {f.name} 失败: {e}")
|
||||
|
||||
if not X_list:
|
||||
print("未找到可用的数据文件")
|
||||
return
|
||||
|
||||
X_all = np.vstack(X_list)
|
||||
|
||||
if standardize:
|
||||
mean = X_all.mean(axis=0)
|
||||
std = X_all.std(axis=0)
|
||||
std[std == 0] = 1.0
|
||||
X_all = (X_all - mean) / std
|
||||
|
||||
if jitter and jitter > 0:
|
||||
rng = np.random.default_rng(random_state)
|
||||
X_all = X_all + rng.normal(scale=jitter, size=X_all.shape)
|
||||
|
||||
reducer = UMAP(
|
||||
n_components=2,
|
||||
n_neighbors=int(max(2, n_neighbors)),
|
||||
min_dist=float(min_dist),
|
||||
spread=float(spread),
|
||||
metric=metric,
|
||||
init=init, # 关键改动：避免谱初始化告警
|
||||
random_state=random_state,
|
||||
)
|
||||
Z_all = reducer.fit_transform(X_all)
|
||||
|
||||
x_min, x_max = float(Z_all[:, 0].min()), float(Z_all[:, 0].max())
|
||||
y_min, y_max = float(Z_all[:, 1].min()), float(Z_all[:, 1].max())
|
||||
pad_x = 0.05 * (x_max - x_min) if x_max > x_min else 1.0
|
||||
pad_y = 0.05 * (y_max - y_min) if y_max > y_min else 1.0
|
||||
|
||||
base_colors = [
|
||||
"#1f77b4","#ff7f0e","#2ca02c","#d62728","#9467bd",
|
||||
"#8c564b","#e377c2","#7f7f7f","#bcbd22","#17becf"
|
||||
]
|
||||
|
||||
start = 0
|
||||
for i, (f, n) in enumerate(zip(paths, counts)):
|
||||
Zi = Z_all[start:start + n]
|
||||
start += n
|
||||
|
||||
fig, ax = plt.subplots(figsize=(6, 5), dpi=150)
|
||||
if context:
|
||||
ax.scatter(Z_all[:, 0], Z_all[:, 1], s=5, c="#cccccc",
|
||||
alpha=0.35, edgecolors="none", label="All")
|
||||
ax.scatter(Zi[:, 0], Zi[:, 1], s=10,
|
||||
c=base_colors[i % len(base_colors)],
|
||||
alpha=0.9, edgecolors="none", label=f.name)
|
||||
|
||||
ax.set_title(
|
||||
f"{f.name} • UMAP(shared) (nn={n_neighbors}, min={min_dist}, metric={metric}, init={init})",
|
||||
fontsize=9
|
||||
)
|
||||
ax.set_xlabel("UMAP-1")
|
||||
ax.set_ylabel("UMAP-2")
|
||||
ax.set_xlim(x_min - pad_x, x_max + pad_x)
|
||||
ax.set_ylim(y_min - pad_y, y_max + pad_y)
|
||||
ax.grid(True, linestyle="--", linewidth=0.3, alpha=0.5)
|
||||
if context:
|
||||
ax.legend(loc="best", fontsize=8, frameon=False)
|
||||
fig.tight_layout()
|
||||
|
||||
out_png = f.with_suffix("").as_posix() + "_umap_shared.png"
|
||||
fig.savefig(out_png)
|
||||
plt.close(fig)
|
||||
print(f"[完成] {f.name} -> {out_png}")
|
||||
|
||||
if make_joint:
|
||||
start = 0
|
||||
fig, ax = plt.subplots(figsize=(7, 6), dpi=150)
|
||||
for i, (f, n) in enumerate(zip(paths, counts)):
|
||||
Zi = Z_all[start:start + n]; start += n
|
||||
ax.scatter(Zi[:, 0], Zi[:, 1], s=8,
|
||||
c=base_colors[i % len(base_colors)],
|
||||
alpha=0.85, edgecolors="none", label=f.name)
|
||||
ax.set_title(f"UMAP(shared) overview (metric={metric}, nn={n_neighbors}, min={min_dist}, init={init})",
|
||||
fontsize=10)
|
||||
ax.set_xlabel("UMAP-1"); ax.set_ylabel("UMAP-2")
|
||||
ax.set_xlim(x_min - pad_x, x_max + pad_x)
|
||||
ax.set_ylim(y_min - pad_y, y_max + pad_y)
|
||||
ax.grid(True, linestyle="--", linewidth=0.3, alpha=0.5)
|
||||
ax.legend(loc="best", fontsize=8, frameon=False, ncol=1)
|
||||
fig.tight_layout()
|
||||
out_png = Path(dir_path) / "umap_shared_overview.png"
|
||||
fig.savefig(out_png.as_posix())
|
||||
plt.close(fig)
|
||||
print(f"[完成] 总览 -> {out_png}")
|
||||
if __name__=="__main__":
|
||||
umap_dir_shared_coords("data")
|
||||
Reference in New Issue
Block a user