如何将映射数据转换为字典,其中每个 XY 坐标都包含一个光谱?

How do I convert mapped data to dictionary, where each XY coordinate contains a spectrum?

我有一个二维地图,其中每个像素都包含一个光谱。我想从这种格式转换数据:

X       Y       Wave        Intensity
-34727.180000   -4204.820000    1.484622    139.193512
-34727.180000   -4204.820000    1.484043    120.991280
-34727.180000   -4204.820000    1.483465    125.905304
-34726.180000   -4204.820000    1.483465    131.262970
-34726.180000   -4204.820000    1.482887    122.784081
-34726.180000   -4204.820000    1.482309    129.853088
-34725.180000   -4204.820000    1.483465    129.655670
-34725.180000   -4204.820000    1.482887    119.567032
-34725.180000   -4204.820000    1.482309    126.097000
-34727.180000   -4203.820000    1.463490    124.331985
-34727.180000   -4203.820000    1.462927    138.189377
-34727.180000   -4203.820000    1.462364    127.824867

到字典,其中键是 X、Y 坐标的元组,值是 3×2 numpy 数组。例如:

DICT = {
    (-34727.180000, -4204.820000): [[1.484622, 139.193512], [1.484043, 120.991280], [1.483465, 125.905304]], 
    (-34726.180000, -4204.820000): [[1.482887, 122.784081], [1.482887, 122.784081], [1.482309, 129.853088]],
    (-34725.180000, -4204.820000): [[1.483465, 129.655670], [1.482887, 119.567032], [1.482309, 126.097000]],
    (-34727.180000, -4203.820000): [[1.463490, 124.331985], [1.462927, 138.189377], [1.462927, 138.189377]]}

这个例子被简化了;我的实际地图包含四个以上的像素(X,Y 坐标),并且每个坐标有 512 个 Wave-Intensity 对。我希望解决方案可以从一个四像素地图推广到一个 400 像素地图,并且每个数组从一个 3×2 numpy 数组推广到一个 512×2 numpy 数组。

最终目标是获取每个坐标的 Wave-Intensity 对,将它们拟合到高斯分布,找到该分布的(最大)振幅,并绘制每个 X、Y 坐标的最大值。这部分问题不需要包含在解决方案中,但是如果有人为这部分问题添加解决方案,那就太好了!

我对不涉及字典的方法持开放态度(例如 4D numpy 数组),但目前我看不到其他方法。随意推荐一种替代方法。目前,我正在使用 pandas:

以原始格式导入数据
import pandas as pd
IN_PATH = r'PATH_TO_FILE'
FNAME = r'\FILENAME.txt'
data = pd.read_csv(IN_PATH+FNAME, sep='\t', skiprows=1)
data.columns = ["X", "Y", "Wave", "Intensity"]

提前致谢!

您可以简单地遍历数据框。但请注意,在您的示例数据中,前几个条目具有相同的 X 和 Y,因此字典条目将被覆盖。

d = {}

for ix, row in df.iterrows():
    d[(row['X'], row['Y'])] = [row[a] for a in row.keys() if not a=='X' and not a=='Y']

编辑: 在同一个键下存储一个像素的所有数据:

d = {}

for ix, row in df.iterrows():
    entry = [row[a] for a in row.keys() if not a=='X' and not a=='Y']
    x,y = row['X'], row['Y']

    if d.get((x,y)):
        d[(x,y)] += [entry]
    else:
        d[(x,y)] = [entry]

第一个pandas.DataFrame.set_index the coordinates, pandas.DataFrame.agg list along axis=1, pandas.DataFrame.groupby the indices, then pandas.DataFrame.groupby.agg to list, and convert pandas.Series.to_dict:

>>> df.set_index(['X', 'Y']).agg(list, 1).groupby(level=(0,1)).agg(list).to_dict()

{(-34727.18, -4204.82): [[1.484622, 139.193512],
                         [1.484043, 120.99128],
                         [1.483465, 125.905304]],
 (-34727.18, -4203.82): [[1.46349, 124.331985],
                         [1.462927, 138.189377],
                         [1.462364, 127.824867]],
 (-34726.18, -4204.82): [[1.483465, 131.26297],
                         [1.482887, 122.784081],
                         [1.482309, 129.853088]],
 (-34725.18, -4204.82): [[1.483465, 129.65567],
                         [1.482887, 119.567032],
                         [1.482309, 126.097]]}

这将在list中给出结果,如果你想要数组,你可以pandas.Series.transformnumpy.array:

>>> df.set_index(['X', 'Y']).agg(list, 1).groupby(level=(0,1)).agg(list).transform(np.array).to_dict()

{(-34727.18, -4204.82): array([[  1.484622, 139.193512],
        [  1.484043, 120.99128 ],
        [  1.483465, 125.905304]]),
 (-34727.18, -4203.82): array([[  1.46349 , 124.331985],
        [  1.462927, 138.189377],
        [  1.462364, 127.824867]]),
 (-34726.18, -4204.82): array([[  1.483465, 131.26297 ],
        [  1.482887, 122.784081],
        [  1.482309, 129.853088]]),
 (-34725.18, -4204.82): array([[  1.483465, 129.65567 ],
        [  1.482887, 119.567032],
        [  1.482309, 126.097   ]])}